![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Departments of Biochemistry and Molecular Biology and Pediatrics, University of Arkansas for Medical Sciences and the Arkansas Children’s Hospital Research Institute, Little Rock, AR 72205
3To whom correspondence should be addressed. E-mail: MockDonaldM{at}uams.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
KEY WORDS: • biotin • avidin • streptavidin • metabolites • urine • europium
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. In these studies, biotin excretion was measured by
first separating biotin from biotin metabolites by HPLC followed by
quantitation using an avidin-binding assay
(ABA).4
However, this two-part assay is labor intensive; direct measurement
in urine without prior chromatography would be desirable. In the
studies presented here, we addressed the following questions concerning
quantitation of biotin in human urine: 1) Does the value
obtained by direct measurement of total avidin-binding substances
agree with biotin quantitated by HPLC followed by ABA in human urine?
2) If not, does the presence of biotin metabolites cause the
disagreement between total avidin-binding substances and true
biotin? 3) If so, can an assay based on streptavidin, which
does not bind biotin metabolites as strongly as avidin, produce a
better estimate of true biotin? |
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The term "total avidin-binding substances" (TABS) in urine is
defined operationally as the value obtained when urine is assayed
directly with an ABA as previously described (2)
. Because
this ABA is a sequential, solid-phase assay, the avidin-binding
substances in urine must bind to the biotin-binding sites on avidin
tightly enough to prevent the avidin from binding to biotinylated
bovine serum albumin (BSA), which is attached in the solid phase. Thus,
the binding must be relatively strong and specific; consequently, TABS
is likely to be restricted mainly to biotin and structurally related
analogs such as the biotin metabolites. This operational definition is
consistent with our previous use of this term (3)
. By
analogy with TABS, here we define "total streptavidin-binding
substances" (TSABS) as the value obtained by direct assay using the
streptavidin-binding assay (SABA).
Both the ABA and the SABA used in these studies measure the ability of
the biotin in urine to occupy all four of the biotin-binding sites
on avidin (or streptavidin) and thus prevent the subsequent binding of
the avidin (or streptavidin) to a biotinylated protein, which has been
immobilized to a solid phase. For both assays, the solid phase is a
plastic microtiter well in a 96-well plate. For the ABA, the
biotinylated protein is BSA; for the SABA the biotinylated protein is
bovine 
-globulin (BIgG). Each assay quantitates the avidin (or
streptavidin) bound by quantitating a linked reporter molecule. For
avidin, the reporter molecule is horseradish peroxidase, and the
detection device is an optical absorbance spectrophotometer. For the
streptavidin, the reporter molecule is europium (Eu), and the detection
system is a time-resolved fluorometer.
Chemicals and reagents.
All chemicals and reagents were analytical grade. Ultrapure deionized water (Millipore, Bedford, MA) was used throughout. D-Biotin, BIgG, biocytin, Tween 20 and dimethyl sulfoxide were purchased from Sigma Chemical (St. Louis, MO). Sulfosuccinimidyl-6-(biotinamido) hexanoate (NHS-LC-Biotin) was purchased from Pierce Chemical (Rockford, IL). Streptavidin labeled with the europium chelate of N1-(p-isothiocyanatobenzyl)-diethylene-triamine-N1,N2,N3,N3-tetraacetic acid (Eu-SA) and Enhancement Solution were purchased from Wallac (Gaithersburg, MD).
Biotin and biotin metabolite standards.
Biotin, bisnorbiotin, biocytin, bisnorbiotin methylketone and biotin
sulfoxide standards were confirmed to be >95% pure using HPLC as
described below. Biotin sulfoxide was prepared from unlabeled biotin
and 3H-biotin by sulfur oxidation in hydrogen peroxide with
confirmation by reverse reduction using sulfhydryl reducing agents as
previously described (4)
. Bisnorbiotin was prepared by
biosynthesis from 14C-biotin, and unlabeled biotin using
Rhodotorula rubra as previously described
(4)
; structure was confirmed by subsequent oxidation to
bisnorbiotin sulfoxide as previously reported (4)
.
Bisnorbiotin methylketone was synthesized and extensively characterized
as reported previously (5)
.
Processing and handling of urine samples.
During collection, urine samples were stored at 4°C. Upon completion,
total volume was measured and an aliquot of the timed collection was
centrifuged at 5000 x g for 10 min at 4°C to
remove cells and debris; samples were then subaliquoted as appropriate
for later analysis and stored at -20°C. Biotin, biocytin,
bisnorbiotin and biotin sulfoxide are stable during handling and
storage up to at least 5 y under these conditions as judged by the
chromatograhic properties of the radiolabeled compounds
(4)
.
HPLC separation of biotin and biotin analogs.
Before analysis, each sample was thawed at 37°C for 30 min, vortexed
and centrifuged at 500 x g for 10 min to remove
cells and debris. Just before HPLC, the pH of each sample was adjusted
to 2.5. Chromatographic fractions were collected and dried as
previously reported (2)
. Each dry sample was reconstituted
in 0.5 mL of water and 0.5 mL of a twofold concentrated solution of the
appropriate assay buffer described below. The samples were serially
diluted in assay buffer and analyzed in quadruplicate by ABA and/or
SABA.
The volume of urine injected on HPLC was adjusted on the basis of
expected biotin concentration. For deficient subjects, the maximum
injected volume was 1 mL, and the resulting injected mass of biotin was
at least 50 fmol. This mass is easily detected despite dilution in HPLC
fractions by both the avidin-binding assay and the
streptavidin-binding assay described below, which each have a
sensitivity of at least 10 fmol injected in a 1-mL sample. Immediately
before injection, the pH of the urine was adjusted to 2.5 using 6 mol/L
HCl, and the sample was filtered through a 0.45-mm (pore size) filter.
The C18 reverse-phase chromatography separates biotin and biotin
analogs on the basis of polarity using a binary gradient system.
Solution A contained 5 x 10-4 L/L
trifluoroacetic acid, pH adjusted to 2.5 with ammonium acetate.
Solution B was an equal volume mixture of acetonitrile and 5 x 10-4 L/L trifluoroacetic acid
(2)
. On the basis of studies using 3H-biotin,
14C-biotin and unlabeled biotin, recovery by HPLC is
>95%, and degradation of biotin during this chromatographic method is
3%.
Measurement of TABS, biotin, and biotin metabolites in urine by ABA.
Biotin was measured by direct ABA at a suitable dilution using the
HEPES buffer as previously described (2)
. For direct
measurement in urine, the dilution was never <1:10, which minimized
the effect of urine pH and salt concentration. Biotin, biocytin,
bisnorbiotin, bisnorbiotin methylketone and biotin sulfoxide were
quantitated against authentic standards in the appropriate HPLC
fractions as previously described (4
,6)
. For these five
compounds, within-run precision is typically 5% and
between-run precision is typically <10%. Absolute standardization
was done gravimetrically for biotin with a cross-check by assaying
3H-biotin of known specific radioactivity. Biocytin was
standardized gravimetrically. Bisnorbiotin and biotin sulfoxide were
standardized radiometrically after synthesis from
14C-biotin and 3H-biotin of known specific
radioactivity as described above. Bisnorbiotin methylketone in human
urine was standardized against biotin with a 0.85 correction factor
(7)
.
Synthesis of biotinylated BIgG.
Biotin was covalently coupled to BIgG according to the modified method
of Diamandis and Christopoulos (8)
. BIgG (10 mg) was
dissolved in 10 mL of 0.3 mol/L sodium bicarbonate, pH 8.5. To this
solution, 0.1 mL of sulfosuccinimidyl-6-(biotinamido) hexanoate
(NHS-LC-biotin, 100 g/L in dimethyl sulfoxide) was added and vortexed.
After incubation for 90 min at room temperature, dialysis at 4°C
against 0.1 mol/L sodium bicarbonate buffer (pH 8.5) containing 154
mmol/L sodium chloride and 14 mmol/L sodium azide was performed to
remove any trace of unreacted NHS-LC-biotin. The biotinylated-BIgG
is stable at -20°C for at least 1 y.
Preparation of biotinylated-BIgG–coated plates.
Biotinyl-BIgG (100 µL; 2 mg/L) in 0.05 mol/L sodium carbonate/bicarbonate buffer, pH 9.6, was added to each well of a 96-well plate (Delfia, Nalge Nunc International, Denmark) and incubated overnight at room temperature. The wells were washed once with water in a 96-multiwell plate washer (EL 403 automated microplate washer; Bio-TeK Corporation, Winooski, VT) to remove unbound protein. To reduce nonspecific binding, plates were countercoated. A 200 µL amount of a 0.1 mol/L NaHCO3 solution, pH 8.5, containing 0.1 g/L BIgG and 7.7 mmol/L NaN3 was added to each well. Plates were incubated for 1 h at room temperature and washed three times with water before use.
Biotin Assay using Eu-SA and time-resolved fluorometry.
For the SABA, instead of the HEPES buffer described previously for the
ABA (2)
, the assay buffer was 50 mmol/L Tris-HCl, pH
7.8, 154 mmol/L NaCl, 7.7 mmol/L NaN3, 0.02 mmol/L
diethylene-triamine pentaacetic acid, 0.5g/L BIgG and 5.0 x 10-4 L/L Tween 20. The assay was conducted
using three incubations. Incubation 1: Samples were prepared as either
an appropriate dilution of a urine sample in assay buffer or an
appropriate dilution of the HPLC fraction containing biotin that had
been reconstituted in assay buffer after drying as described previously
(2)
. The sample (100 µL) was added to an
uncoated U-bottomed multiwell plate (Dynatech Laboratories, McLean,
VA). A 50-µL solution of 83 nmol/L Eu-SA in assay
buffer was added to each well with mixing. Incubation for 45 min at
room temperature with shaking (Plateshaker, 1296–024 Wallac) allowed
the free biotin in the sample to bind to the tetrameric Eu-SA.
Published pharmacokinetic data and our empirical observations both
indicate that this reaction reached equilibrium within a few minutes.
In parallel, a set of known biotin concentrations was used to construct
a standard curve.
Incubation 2.
An aliquot of 75 µL of Incubation 1 was transferred from each well to the corresponding well in a flat-bottomed 96-well plate that had been previously coated with biotinylated-BIgG as described above. During an overnight incubation at room temperature, any Eu-SA molecules with at least one unoccupied biotin-binding site will bind to the solid phase. The more biotin present in the unknown, the fewer Eu-SA molecules that bind to the coated plate in Incubation 2. The second plate was washed six times with water to remove unbound and nonspecifically bound Eu-SA.
Incubation 3.
The amount of Eu bound to the plate was quantitated by acid
denaturation (which releases the Eu) followed by chelation (which
provides an appropriate environment for fluorescence). Both
denaturation and chelation were accomplished by the enhancement
solution which contained (per L) 15 µmol of
2-naphthoyltrifluoracetone, 50 µmol of
tri-n-octyl-phosphineoxide, 1.0 g of Triton X-100,
100 mmol of acetic acid and 6.8 mmol of potassium hydrogen phthalate.
The plate was developed by incubation with 200 µL of
enhancement solution for 60 min at room temperature with shaking. For
reading of an individual well, a 100-µs light burst
was followed by a "black out" period of 400 µs,
which greatly reduced fluorescence from the shorter-lived
biological fluors. Next, fluorescent photons were counted for 400
µs using a single-photon counting,
time-resolved fluorometer (1234 Delfia Wallac) equipped with a 320
nm excitation filter and 615 nm emission filter. This pair is
appropriate for Eu3+ fluorescence. This cycle was repeated
1000 times, resulting in an analysis time for each sample well of
1 s. Results are reported as the mean of fluorescent photons counted
per second (cps). Analytic variation is reported here as the standard
deviation among replicate wells (typically n = 3 or
4) and is depicted in the figures as ±1 SD error bars.
Human research.
Protocols for urine collection were approved by the University of
Arkansas for Medical Science’s Human Research Advisory Committee and
the University of Iowa Institutional Review Board. To examine samples
over a broad range of biotin nutritional status, urine samples were
obtained from the following sources: 1) Samples
(n = 28) from four healthy individuals (two women)
who were made progressively (but asymptomatically) biotin deficient
with 3 wk of egg-white feeding (1)
. 2)
Samples (n = 68) from six healthy individuals (six
women) who were initially supplemented with 300 µg
(1.23 µmol) of biotin for 5–14 d followed by a
washout of 7 d and then by 4 wk of egg-white feeding, followed
by 2 wk of biotin supplementation at 30 µg/d (0.12
µmol). 3) Samples (n
= 42) from 16 healthy women who became spontaneously marginally
biotin deficient during pregnancy and from five healthy control women
who were not biotin deficient. 4) Samples
(n = 29) from three healthy individuals (two women)
who were made progressively, but asymptomatically biotin deficient with
3 wk of avidin feeding.
Statistical methods.
Significance of correlation between groups (e.g., between assay results) was tested by simple linear regression (StatView 5.01, SAS Institute, Cary, NC). The significance level was chosen at P < 0.05. Whether the regression line differed significantly from the line of identity was judged in two ways, i.e., whether the 95% confidence limits for the slope included 1.0 and whether the 95% confidence limits for the y-intercept included 0.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, a moderately strong and significant correlation between TABS and true
biotin was detected (r = 0.78), and the regression
coefficient (slope) was significantly different from zero at
P < 0.0001. The equation for the linear regression was
as follows: TABS (nmol/L) = 1.1 · biotin (nmol/L) + 19.6
nmol/L.
|
,
we concluded that a component of the measurement of TABS was indeed
biotin and that the biotin component was measured accurately. However,
the lower 95% confidence interval (CI) for the intercept did not
include zero. This finding suggested that one or more additional
factors contributed to the TABS value. When excretion rates for TABS and true biotin were calculated as nmol/24 h, the correlation remained strong (r = 0.82, data not shown). For excretion rates, the regression coefficient again included the line of identity within the 95% confidence limits, but again the 95% confidence limits of the intercept did not include zero.
To further characterize the error, we calculated the number of samples for which the TABS value was either greater than or less than the true biotin value, allowing for the likely analytical error. For our estimate of analytical error, we chose twice the intraday CV value (CVintra) for the ABA. Given that CVintra = ± 10%, the 95% confidence limits for analytical error = ± 20%. Only five of the 133 TABS values underestimated the true biotin concentration by more than the analytical error; in contrast, 112 TABS values exceeded the biotin value by more than the analytical error.
We investigated the potential role of biotin metabolites as the cause
of the overestimate. In individuals supplemented with biotin, the
proportion (mol/100 mol) of total urinary biotin plus metabolites
attributable to the metabolites decreased from about half
(4)
to about one fifth (9)
. Accordingly, we
hypothesized that if the direct ABA errors arose primarily from the
metabolites, the errors would be smaller in samples from supplemented
individuals than in those from normal individuals. We subgrouped the
133 samples according to whether their 24-h biotin excretion was
greater than the upper limit of normal ("supplemented"), within the
normal range ("normal") or less than the lower limit of normal
("deficient"). These data are depicted in Figure 1A
, B
and C
. The 95% CI for the slope from the supplemented group included 1
(Table 1)
. In contrast, the slope of the regression lines for both the
normal and deficient subgroups was substantially and significantly
>1.0; the 95% CI did not include a slope of 1. Moreover, the slope
from the deficient subjects was even greater than that of the normal
subjects. These observations were all consistent with the hypothesis
that biotin metabolites contributed substantially to TABS.
|
and presented in Table 1
, the overestimate in TABS correlated
positively, significantly (P < 0.0001) and strongly
with the concentration of biotin metabolites. (For this analysis, we
excluded a single outlier with error of more than twice that for any
other sample and >3 SD of the group.)
|
Solid phase to capture streptavidin.
The degree of biotinylation of BIgG was assessed by measuring the amount of Eu-SA that bound to a well coated with biotinyl-BIgG as follows. Wells in a suitable multiwell plastic plate were coated with either biotinyl-BIgG or native BIgG. Serial dilutions of Eu-SA solution were added in triplicate to the coated wells and incubated for 30 min. After being washed with water, time-resolved fluorescence intensity was quantitated for each well as described in Methods. Wells coated with biotinyl-BIgG reproducibly bound at least 50 times more Eu-SA than wells coated with native BIgG.
Sensitivity.
A standard curve was constructed using known concentrations of biotin.
As depicted in Figure 3
, the standard curves for the SABA and the ABA were quite similar for
biotin. The curve shapes diverged slightly at biotin concentrations
>30 pmol/L. Relative sensitivity for the two curves was similar. If
sensitivity is defined as the first biotin concentration in a standard
curve that produces a signal significantly different from zero as
judged by one-way ANOVA with Dunnett’s post-hoc test,
sensitivity was 8 pmol/L for the SABA. For the ABA, sensitivity was 12
pmol/L. Alternatively, if sensitivity is defined as the concentration
of biotin whose signal is three standard deviations different from the
value at zero biotin, the sensitivity of the SABA was 5 pmol/L, and the
sensitivity of the ABA was 8 pmol/L. Thus, the overall sensitivity of
the assays were similar.
|
, bisnorbiotin was undetectable by the SABA. Biotin
sulfoxide was detectable with an efficiency of only 10% compared
with biotin. Bisnorbiotin methylketone was separated from other
(strept)avidin binding substances by HPLC and assayed by both SABA and
ABA. Bisnorbiotin methylketone was undetectable by the SABA (i.e.,
<5% compared with ABA). In contrast, biocytin detectability by the
SABA was the same as biotin. Precision.
The within-run precision of the SABA was determined by replicate analysis of five urine samples without HPLC separation. The urine samples were chosen to have TSABS concentrations varying from low normal to high normal. The CV ranged from 2.8 to 5.6%; the mean was 4.5%. Between-run precision was determined from triplicate measurements on one urine sample assayed on five different days. The average CV was 11%.
Linearity of dilution.
The linearity of the assay was assessed by diluting a urine sample containing a large concentration of biotin with assay buffer. The TABS concentrations ranged from 46 to 230 nmol/L. These values were plotted against biotin values calculated from the dilution factor. Agreement between observed and expected values was excellent; slope = 1.02 ± 0.05, y-intercept = -2 ± 5, and r = 0.999.
Validation.
Measurement of biotin in HPLC fractions by SABA was next validated
against the ABA of the same HPLC fractions. Samples (n
= 29; 10 deficient, 10 normal and 9 supplemented) were assayed.
The assays agreed within analytical error (Fig. 4
). The slope and the intercept of the regression line (Table 2
) were not significantly different from those of the line of identity.
These observations provide evidence that each assay measured biotin
with approximately the same accuracy and precision; further, HPLC
separates biotin from its metabolites and from any substances that
interfere with the SABA.
|
|
To assess the performance of the SABA in measuring biotin directly, we
assayed a subgroup of 92 samples. This subgroup was chosen from ongoing
studies in which true biotin had been measured by HPLC with ABA. TSABS
correlated moderately strongly and highly significantly with true
biotin (Table 2)
. However, the upper 95% confidence limit for the
slope of the regression line fell below 1, and the lower 95%
confidence limit for the intercept did not reach the origin.
We further characterized the errors by comparing the TSABS values to the upper and lower limits of analytical error. For 13 of the 92 samples, TSABS were less than the lower limit of analytical error. In contrast, for 62 of the 92 samples, TSABS were greater than the upper limit of analytical error.
We further investigated the TSABS errors by examining the correlation
between TSABS and biotin in the three nutritional status subgroups. As
shown in Figure 5
, we observed the expected decrease in the strength and significance of
the correlations due to the smaller number of samples (Table 2)
,
especially in the supplemented group, which had the smallest number of
samples. For each nutritional subgroup, the 95% confidence limits for
slope and intercept expanded substantially, thereby encompassing a
slope of 1 and an intercept through the origin.
|
, suggesting that biotin
metabolites are not primarily responsible for the TSABS overestimates.
In a subset of 62 samples from six individuals that spanned the range
of biotin status from deficient to supplemented, we measured TABS and
TSABS directly with the appropriate assay. Although statistically
significant, the correlation was weak (r = 0.69) (Table 2)
. Consistent with our hypothesis that biotin metabolites contribute
substantially to the overestimate in TABS, the slope of the regression
line was significantly less than 1 (Table 2)
.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The correlation coefficient between the TABS errors (=TABS -
true biotin) and total metabolite concentration (=bisnorbiotin + biotin
sulfoxide + bisnorbiotin methylketone + biocytin) is 0.80 and the
slope of the regression line is 0.81; these findings suggest that
80% of the overestimate can be attributed to biotin metabolites.
However, the 95% CI for slope do not quite incorporate the line of
identity. Thus, these observations suggest that biotin metabolites are
the primary source of the overestimate of biotin by direct ABA, but may
not be the only source of error.
An artifact attributable to biotin metabolites is consistent with the following published observations:
1) As early as 1963, N. M. Green established that
structural analogs of biotin such as the biotin catabolites and
biocytin exhibit strong binding to avidin (10)
, although
the equilibrium binding constants are often substantially smaller than
that of biotin itself.
2) Various biotin metabolites are present in human urine in
substantial quantities. For example, in a study of six healthy adults
not receiving biotin supplements, bisnorbiotin accounted for a mean
value of 52% of the total of biotin plus biotin metabolites, whereas
biotin accounted for only 32%. Bisnorbiotin methylketone accounted for
8%, biotin sulfone for 3.6% and biotin sulfoxide 4%
(7
,11
,12)
. Biocytin contributes 10% of the molar total
(13)
.
3) The mole percent attributable to a given metabolite
varies substantially among individuals within a given nutritional
status, i.e., deficient, normal and supplemented (4)
, and
also varies with biotin status (1
,9)
.
4) The proportion attributable to biotin metabolites is
substantially increased in certain common situations such as pregnancy
and therapy with anticonvulsants (9
,14
15
16
17
18)
.
These observations provide evidence that biotin metabolites are present in large enough quantities to produce the error seen. Because the metabolite profile varies from individual to individual, metabolites are capable of producing variation in error observed among individuals with normal biotin status. Moreover, the relative proportions of metabolites varies with biotin nutritional status in a way consistent with the observed variation in the magnitude of the direct ABA error; the proportion of metabolites and the magnitude of error decrease with increasing biotin nutritional status and the resulting relatively selective increase of biotin excretion.
Having identified biotin metabolites as a substantial source of error
in the direct assay of biotin in human urine using an ABA, we
hypothesized that an assay based on a molecule that binds biotin
metabolites less tightly than avidin would give more accurate results
in a direct assay. As documented in the validation studies above, the
biotin catabolites were much less detectable than biotin in the SABA.
Bisnorbiotin and bisnorbiotin methylketone were virtually undetectable
in the SABA. Moreover, the combination of the relatively small molar
contribution of biotin sulfoxide (5–10%) and its limited
detectability (10% relative to biotin) suggests that biotin sulfoxide
will make a negligible contribution to TSABS. In contrast, the ratio of
biocytin to biotin in urine is 1:10. Because it is as detectable as
biotin, biocytin will contribute 10% to TSABS.
Despite this desirable metabolite selectivity, the SABA did not produce satisfactory accuracy in the quantitation of biotin when applied directly to human urine. The error in TSABS exceeded likely simple analytical error in 62 of the 92 samples assayed. Given that an analytical range of ± 22% was examined, the overestimate in TSABS cannot be attributed to biocytin. The studies characterizing the nature of the TSABS error suggest that the etiology of the error is different from that of the ABA. The intercept of the plot of TSABS vs. biotin was significantly greater than zero suggesting the presence of an interfering substance (or substances) not related to biotin. This impression was strengthened by examining the relationship of the error (TSABS - biotin) to the total concentration of metabolites. We speculate that rare earth elements in whole urine are the interfering substances. They exhibit long-lived fluorescence. Our studies of commercially available bovine serum albumin, human serum albumin and bovine immunoglobin indicate that these proteins are contaminated with amounts of rare earth elements that produce a large nonspecific signal in the time-resolved fluorescent assay. Although the amounts of albumin and other proteins in urine are small compared with their concentrations in blood, the quantities of protein appear to be sufficient to produce the observed artifacts. Moreover, rare earth metals not bound to proteins provide an additional but unpredictable source of error. To the extent that they bind nonspecifically to the BSA-coated plate, free rare earth elements would not be washed away in the SABA. On the basis of studies using EDTA and dialysis, this type of reversible binding does occur. Finally, the excellent agreement between the SABA and the ABA when applied to the HPLC fractions containing biotin provides strong evidence that the interfering substances are removed from the biotin fraction by HPLC, as would be predicted on the basis of the polarity characteristics of both free rare earth cations and rare earth metals bound to protein. Further investigation utilizing metal ion chelators to prevent the hypothesized binding of free rare earth elements to the protein-coated plate and utilization of ultrafiltration to remove the hypothesized rare earth element:protein complexes could be used to test this speculation; such studies could conceivably lead to sample preparation techniques that would be less labor intensive than HPLC and yet yield an accurate quantitation of biotin.
Theoretically, any substance that interferes with either the binding of biotin to the four biotin-binding sites that are present on avidin and streptavidin or the binding to the solid phase of avidin or streptavidin molecules that have at least one unoccupied biotin-binding site will produce overestimates and underestimates, respectively. Although such mechanisms remain theoretical possibilities, our previous studies with a variety of structural analogs of the biotin molecule indicate that alterations beyond very modest changes, such as those seen in biotin catabolites, produce compounds that cannot effectively compete with biotin for binding to avidin (and presumably streptavidin) despite being present at concentrations several orders of magnitude greater than biotin.
In summary, the studies described here provide important new insight into measuring biotin in human urine. The error in direct ABA is more fully characterized than previously, and the source of the error is primarily biotin metabolites. The development and validation of a SABA based on time-resolved fluorescence is described. Although the SABA is as sensitive and as accurate as the ABA, our studies indicate that the SABA is not satisfactory for direct quantitation of biotin in human urine.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Sponsored by RO1 DK36823 National Institutes of Health, NIDDK, MO1 RR14288 National Institutes of Health, NCRR and M01 RR00059 National Institutes of Health, NCRR.
4 Abbreviations used: ABA, avidin-binding assay; BIgG, bovine -globulin; BSA, bovine serum albumin; CI, confidence interval; Eu-SA, europium-streptavidin; SABA, streptavidin-binding assay; TABS, total avidin-binding substances; TSABS, total streptavidin-binding substances.
Manuscript received January 31, 2001. Initial review completed April 6, 2001. Revision accepted May 21, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1.
Mock N., Malik M., Stumbo P., Bishop W., Mock D. Increased urinary excretion of 3-hydroxyisovaleric acid and decreased urinary excretion of biotin are sensitive early indicators of decreased status in experimental biotin deficiency. Am. J. Clin. Nutr. 1997;65:951-958
2. Mock D. M. Determinations of biotin in biological fluids. McCormick D. B. Suttie J. W. Wagner C. eds. Methods in Enzymology 1997;279, Part I:265-275 Academic Press New York, NY. [Medline]
3. Zempleni J., Mock D. M. Biotin. Song W. O. Beecher G. R. eds. Modern Analytical Methodologies on Fat and Water-Soluble Vitamins 2001:459-466 Wiley Baltimore, MD.
4. Mock D. M., Lankford G. L., Cazin J., Jr Biotin and biotin analogs in human urine: biotin accounts for only half of the total. J. Nutr. 1993;123:1844-1851
5.
Kazarinoff M. N., Im W.B., Roth J. A., McCormick D. B., Wright L. D. Bacterial degradation of biotin VI. Isolation and Identification of ß-hydroxy and ß-keto compounds. J. Biol. Chem. 1972;247:75-83
6.
Zempleni J., McCormick D. B., Mock D. M. Identification of biotin sulfone, bisnorbiotin methyl ketone, and tetranorbiotin-l-sulfoxide in human urine. Am. J. Clin. Nutr. 1997;65:508-511
7. Zempleni J., Mock D. M. Advanced analysis of biotin metabolites in body fluids allows a more accurate measurement of biotin bioavailability and metabolism in humans. J. Nutr. 1999;129:494S-497S
8.
Diamandis E. P., Christopoulos T. K. The biotin-(strept)avidin system: principles and applications in biotechnology. Clin. Chem. 1991;37(5):625-628
9.
Mock D. M., Heird G. M. Urinary biotin analogs increase in humans during chronic supplementation: The analogs are biotin metabolites. Am. J. Physiol. 1997;272:E83-E87
10. Green N. M. Avidin 3. The nature of the biotin-binding site. Biochem. J. 1963;89:599[Medline]
11. Mock D. M. Introduction to symposium: nutrition, biochemistry and molecular biology of biotin. J. Nutr. 1999;129:476S
12. Mock D. M. Biotin status: which are valid indicators and how do we know?. J. Nutr 1999;129:498S-503S
13. Wolf B., Norrgard K., Pomponio R. J., Mock D. M., Secor-McVoy J., Fleischhauer K., Shapiro S., Blitzer M. G., Hymes J. Profound biotinidase deficiency in two asymptomatic adults. Am. J. Med. Genet. 1997;73:5-9[Medline]
14. Mock D. M., Dyken M. E. Biotin deficiency results from long-term therapy with anticonvulsants. Gastroenterology 1995;108:A740(abs.)
15. Wang K.-S., Patel A., Mock D. M. The metabolite profile of radioisotope labeled biotin in rats indicates that biotin metabolism is similar to that in humans. J. Nutr. 1996;126:1852-1857
16. Mock D. M., Stadler D. D. Conflicting indicators of biotin status from a cross-sectional study of normal pregnancy. J. Am. Coll. Nutr. 1997;16:252-257[Abstract]
17.
Mock D. M., Stadler D., Stratton S., Mock N. I. Biotin status assessed longitudinally in pregnant women. J. Nutr. 1997;127:710-716
18.
Mock D. M., Dyken M. E. Biotin catabolism is accelerated in adults receiving long-term therapy with anticonvulsants. Neurology 1997;49:1444-1447
This article has been cited by other articles:
|
|
 |
|
  A. Bogusiewicz, S. L Stratton, D. A Ellison, and D. M Mock Biotin accounts for less than half of all biotin and biotin metabolites in the cerebrospinal fluid of children Am. J. Clinical Nutrition, November 1, 2008; 88(5): 1291 - 1296. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. E. Peterson, L. R. McDowell, R. J. McMahon, N. S. Wilkinson, O. Rosendo, W. M. Seymour, P. R. Henry, F. G. Martin, and J. K. Shearer Balance and serum concentration of biotin in sheep fed alfalfa meal-based diets with increasing level of concentrate J Anim Sci, April 1, 2004; 82(4): 1165 - 1169. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
