Impaired Biotinidase Activity Disrupts Holocarboxylase Synthetase Expression in Late Onset Multiple Carboxylase Deficiency*

Biotinidase catalyzes the hydrolysis of the vitamin biotin from proteolytically degraded biotin-dependent carboxylases. This key reaction makes the biotin available for reutilization in the biotinylation of newly synthesized apocarboxylases. This latter reaction is catalyzed by holocarboxylase synthetase (HCS) via synthesis of 5′-biotinyl-AMP (B-AMP) from biotin and ATP, followed by transfer of the biotin to a specific lysine residue of the apocarboxylase substrate. In addition to carboxylase activation, B-AMP is also a key regulatory molecule in the transcription of genes encoding apocarboxylases and HCS itself. In humans, genetic deficiency of HCS or biotinidase results in the life-threatening disorder biotin-responsive multiple carboxylase deficiency, characterized by a reduction in the activities of all biotin-dependent carboxylases. Although the clinical manifestations of both disorders are similar, they differ in some unique neurological characteristics whose origin is not fully understood. In this study, we show that biotinidase deficiency not only reduces net carboxylase biotinylation, but it also impairs the expression of carboxylases and HCS by interfering with the B-AMP-dependent mechanism of transcription control. We propose that biotinidase-deficient patients may develop a secondary HCS deficiency disrupting the altruistic tissue-specific biotin allocation mechanism that protects brain metabolism during biotin starvation.

Because of the importance of biotin in cell metabolism, higher organisms face a constant threat to their survival because they are incapable of synthesizing the vitamin. The situation is further complicated by the limited availability of biotin in nature, most of which is protein-bound and not directly accessible (2,14). During evolution, mammals developed what is known as the biotin cycle, which allows them to cope with the low availability of this critical nutrient (Fig. 1). This system depends on two enzymes; holocarboxylase synthetase (HCS) and biotinidase (1,2). HCS is responsible for the activation, via biotinylation, of all biotin-dependent carboxylases in human cells. The process takes place in a two-step, ATPdependent reaction in which biotin is first activated to 5Ј-biotinyl-AMP (B-AMP) and then transferred to a specific and highly conserved lysine residue in all biotin-dependent carboxylases (2,15,16). Biotinidase catalyzes the release of biotin from biotinylated peptides or biocytin (biotinyl-lysine), products generated by intestinal digestion of nutrient proteins or during carboxylase turnover (endogenous biotin recycling) (2,17).
We recently showed that HCS is an obligate participant in biotin-mediated transcriptional regulation (Fig. 1). The underlying mechanism requires B-AMP, the product of the HCS reaction, which activates a signal transduction cascade involving soluble guanylate cyclase (sGC) and cGMP-dependent protein kinase (PKG) (4,7). In the presence of biotin, the HCS-sGC-PKG pathway induces the expression of the components of the biotin cycle required for its transport and utilization: the sodium-dependent multivitamin transporter, PC, PCC, MCC, ACC, and HCS (3,5,13). Paradoxically, biotin deficiency results * This work was supported by Consejo Nacional de Ciencia y Tecnología Grant 48862 and Programa de Apoyo a Proyectos de Investigació n e Innovació n Tecnoló gica Grant IN220206-3 and funds from the Universidad Nacional Autó noma de Mé xico. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Recipients of doctoral scholarships from the Consejo Nacional de Ciencia y Tecnología. 2  in reduced expression of these genes in tissues such as liver, kidney, and muscle but not brain. Although this would seem to be contrary to the need for scavenging biotin during limited supply, we showed that this pattern of gene repression is an altruistic tissue-specific contingency mechanism that, by down-regulating biotin utilization in selected tissues, allows a concerted supply of the remaining vitamin to the brain (3). In this organ, PC plays two essential roles: as a key player in anaplerosis of the Krebs cycle through pyruvate carboxylation and in the restoration of ␣-ketoglutarate lost during the release of glutamate and ␥-aminobutyric acid from neurons and glutamine export from glia (18,19). In humans, the biotin cycle can be disrupted by genetic deficiency of holocarboxylase synthetase (HCS deficiency (MIM 253270)) or biotinidase (BTD deficiency (MIM 253260)), resulting in neonatal or juvenile onset forms, respectively, of the disease multiple carboxylase deficiency (MCD) (2,20,21). Although the two diseases differ in the age of onset of symptoms, they share a number of clinical and biochemical manifestations, including decreased activities of all carboxylases, organic acidemia, hyperammonemia, dermatitis, alopecia, seizures, and coma. In biotinidase-deficient patients, neurological damage may also include mental retardation, hearing loss, and optic nerve atrophy (2). Although potentially lethal, most of clinical and biochemical manifestations of neonatal and juvenile MCD can be successfully treated with pharmacological doses of biotin.
The biotin-responsiveness of neonatal MCD patients is associated primarily with having at least one allele expressing a mutant HCS with an elevated K m for biotin, which allows for increased functional activity at high concentrations of biotin (1)(2)(3)(4)(5). However, based on the participation of HCS in the biotindependent transcriptional regulation of the biotin cycle, we have suggested that the clinical and biochemical deficits in HCS-deficient patients reflect the combined effects of the low affinity of the mutant enzyme for biotin and the concomitant reduction in carboxylase and HCS mRNA levels (3,4). In biotinidase-deficient patients, the biotin cycle is largely intact because free biotin can be successfully utilized for the biotinylation of carboxylases. Here the deficit has been thought to be in the inadequacy of the biotin supply because of the inability to recycle proteinbound biotin from endogenous or nutrient sources (2). Therefore, supplementation with biotin at pharmacologic doses is thought to compensate for the loss of access to the additional biotin that would normally be available from protein sources (2). Although the ultimate consequence of biotinidase deficiency is the interruption of the metabolic pathways where biotindependent carboxylases participate, the clinical manifestations that distinguish this disorder from HCS deficiency, especially in relation to neurological presentation, are not fully understood (22).
In this work, we use fibroblasts from a biotinidase-deficient patient as an experimental model to study the role of this enzyme in carboxylase biotinylation and in HCS-sGC-PKG-dependent expression of biotin-dependent carboxylases and HCS under conditions of biotin deficiency and supplementation. Our results show that in biotinidase-deficient cells, biotin starvation results in a more rapid reduction in carboxylase biotinylation and in the expression of PC and MCC than in normal fibroblasts. We also demonstrate that in biotinidase-deficient cells the expression and activity of HCS is lower than in control cells. We propose that, in the absence of biotin supplementation, biotinidase-deficient patients may develop a secondary HCS deficiency that, combined with the primary biotinidase deficiency, may disrupt the altruistic regulation of biotin utilization that protects brain metabolism against vitamin starvation.
Cell Culture Experiments-The methods for biotin starvation of cell cultures are essentially as described previously (3,4). Briefly, the cells were grown in biotin-replete or biotin-deficient medium at 37°C with 5% CO 2 for up to 13 days. The medium was changed at 3-day intervals. For carboxylase biotinylation recovery experiments, the cells grown in biotin-deficient medium were stimulated with biotin or biocytin, at concentrations from 1 to 100 nM for 24 h.
Western Blot Analysis for HCS Expression and Carboxylase Biotin Content-Crude extracts from human cell cultures (30 -100 g of total protein) were fractionated by SDS-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore). The membranes were incubated in a 1:3000 solution of rabbit HCS antibody or with a 1:2000 solution of streptavidin-AP conjugate (Roche Applied Science) or with a 1:500 solution of goat MCC or PCC antibodies (Santa Cruz Biotechnology). Visualization of HCS bands was performed using a BM chemiluminescence Western blotting kit (Roche Applied Science). Biotin-containing bands were quantitated using an FX image analyzer (Bio-Rad) as described above. The protein concentration used in these experiments was determined using a ND-1000 spectrophotometer (Nanodrop Technologies, Inc.), and confirmation of equal amounts of total protein in every lane was done by staining the gels with Coomassie Blue before protein transfer to PVDF membranes.
Effect of Biotin and cGMP on Carboxylase Expression in Biotin-starved Cells-To determine the involvement of sGC on the recovery of biotinylation and carboxylase protein levels, biotinstarved normal fibroblasts were treated with 50 M ODQ, a specific inhibitor of sGC, for 3 h (4). After this period, 1 M biotin was added to the medium for 48 or 72 h, and the effect on MCC protein levels was compared with biotin-deficient cells stimulated by biotin without ODQ and cells grown continuously in normal medium (control cells). Alternatively, MCD-BD cells grown in biotin-free medium for 13 days were stimulated with 1 M biotin or 1 mM 8-Br-cGMP, a nonhydrolyzable analogue of cGMP. The cells were harvested after 48 or 72 h, and the MCC protein levels were determined as described above.
Reverse Transcription-PCR-Procedures for RNA isolation, cDNA synthesis, and PCR have been previously described (3,4). 5 g of total RNA and 0.5 M gene-specific oligonucleotide primers were used for cDNA synthesis and 0.3 M specific sense and antisense primers were used to give 200 -300 bp of PCR products. The oligonucleotides used to amplify human mRNA were: HCS: 5Ј-CCC GAG CTC CGT CTC CTG GAT CGG-3Ј and 5Ј-CCC AAG CCT TTT ACC GCC GTT TGG GGA-3Ј (T m ϭ 58°C); Biotinidase, 5Ј-ATC TAT GAA CAG CAA GTG ATG ACT-3Ј (T m ϭ 66°C) and 5Ј-AGG GAC CAG GGT GAA ATT GTC ATA-3Ј (T m ϭ 70°C); ␤-actin: 5Ј-GGG TCA GAA TTC CTA TG-3Ј and 5Ј-GGT CTC AAA CAT GAT CTG GG-3Ј (T m ϭ 58°C). PCR products were separated on 1% agarose gels and stained with ethidium bromide. The amount of PCR product was determined by densitometry by using a Fluor-S-imager (Bio-Rad) as previously described (3,4). The procedure was validated in prior studies by PCR amplification of different concentrations of cDNA fragments of HCS, biotinidase, and ␤-actin (data not shown). The number of PCR cycles was also varied and plotted against fluorescence intensity to ensure that experiments were done within the exponential phase. For every experiment, the constitutive ␤-actin mRNA was used as the reference cellular transcript. It was present at equivalent levels in all RNA samples.
HCS Activity Assay Using p67 as Biotinylation Substrate-To determine HCS activity in normal and MCD-BD fibroblasts, we used a modification of the protocol described previously (16,25). Briefly, a pFLAG vector (Sigma) containing a cDNA fragment encoding the last 67 amino acids (640 -703) of the ␣ subunit of human PCC was used to transform wild type E. coli XL1 Blue and E. coli C-124, a mutant strain unable to synthesize dethiobiotin, an essential intermediate in the production of biotin. Log phase XL-1 and C-124 cultures in L-broth medium were transferred to a biotin-free medium (7.5 mM (NH 4 ) 2 SO 4 , 33 mM KH 2 PO 4 , 60 mM K 2 HPO 4 , 1.7 mM sodium citrate, 1 mM MgSO 4 , 0.2% dextrose, 0.1% casamino acids) and 2 mM isopropyl ␤-D-thiogalactopyranoside and incubated at 37°C for 4 h. The cells were sonicated three times with 10-s pulses and centrifuged at 15,000 rpm for 20 min. The proteins in the supernatant were resolved by 12% acrylamide gel electrophoresis (100 g of total protein/lane) and transferred to a PVDF membrane. Two biotinylated proteins are possible in cells expressing p67: p67, at 14 kDa, and BCCP, the 18-kDa subunit of E. coli ACC. To detect their positions on the gel, one lane containing proteins expressed by XL1 cells transformed with p67 was cut off from the membrane and incubated with streptavidin-AP to detect the biotinylated proteins (see Fig. 6A). The lower band, corresponding to p67, was used as a reference to cut out the section of the membrane in adjacent lanes containing unbiotinylated or apo-p67 expressed by E. coli C-124. The membrane pieces containing apo-p67 were used, in solid phase format, for HCS assays. HCS activity was monitored by incubating the membrane pieces for 1 h at 25°C in 150 l of reaction buffer containing Tri-HCl, pH 8.0, 50 mM reduced glutathione, 22.5 mM MgCl 2 , 5 mM ATP, 1-3 Ci of [ 3 H]biotin, and 100 g of total protein of crude extracts from normal or MCD-BD fibroblasts. For these experiments cells grown in biotin-supplemented medium were preincubated for 6 h with 1 M nonradioactive biotin and 63 M cycloheximide to block free biotinylation sites and prevent de novo carboxylases synthesis (5). Crude extracts were prepared as previously described and passed twice through an Amicon ultra centrifugal device (Millipore) to eliminate nonradioactive biotin. The radioactive biotin incorporated into the membrane-bound p67 was estimated using a Beckman LS 6500 scintillation counter. Under these conditions, the p67 was in excess, and the assay was linear for the 1 h of incubation.
cDNA and Genomic DNA Sequencing-To determine the mutations responsible for the phenotype of MCD-BD cells, the biotinidase cDNA was cloned in the pGEM vector (Promega). Biotinidase exons were amplified from genomic DNA as previously described (26,27) and subcloned also in pGEM. Both cDNA and exons were sequenced at Laragen (Los Angeles, CA).
Statistical Analysis-All of the experiments were done in triplicate and at least three different times. The results of biotin starvation on mRNA were normalized to ␤-actin mRNA and expressed as a percentage of mRNA levels observed in cells grown in biotin-replete medium. The data are presented as the mean of three different experiments Ϯ S.E. unless otherwise indicated. Statistical significance of p67 biotinylation results obtained with normal or MCD-BD cells were analyzed at 0.05 and 0.01 levels of significance using Student's t test one-way ANOVA.

Molecular and Functional
Characterization of the Biotinidase-deficient Cell Line MCD-BD-To characterize the cell line MCD-BD used as an experimental model in this study, we first identified the mutations responsible for biotinidase deficiency by sequencing the cDNA encoding this enzyme. This procedure resulted in the identification of a transversion 1330G 3 C, which causes a substitution of His for Asp 444 (D444H), and a single base transition 511G 3 A, resulting in a substitution of Thr for Ala 171 (A171T) (Fig. 2A). These mutations have been previously reported (26) and are considered a common cause of profound biotinidase deficiency in children ascertained by newborn screening in the United States (28). Because these mutations were originally described as a double mutation allele (28), we sequenced all four biotinidase exons from genomic DNA to search for another mutation. This procedure confirmed the identified mutations and did not reveal the presence of any other mutation.
To verify the impact of these mutations, we tested the biotinylation status of the carboxylases PC, PCC, and MCC in MCD-BD cells and compared the results with normal fibroblasts (positive control) and MCD-MK fibroblasts (negative control). The latter cells have unbiotinylated carboxylases in the standard biotin-replete medium caused by a homozygous, high K m R508W mutation in HCS but are restored to normal in medium containing 100ϫ biotin (23). The cells were grown in biotin-replete medium to 80% confluence, and total proteins from cell extracts were resolved by polyacrylamide gel electrophoresis. The biotin content in carboxylases was visualized by Western blot using streptavidin-AP. Three bands were identified in normal cells corresponding to PC (128 kDa), MCC-␣ subunit (76 kDa), and PCC-␣ subunit (72 kDa). As we previously reported, ACC (265 kDa) is not detected using this colorimetric assay (3). PCC and MCC appear as a broad band or doublet, clearly distinguishable from the fainter PC band (Fig. 2B). In contrast, MCD-BD cells exhibited reduced MCC-PCC biotinylation in biotin-replete medium, to approximately two-thirds of the levels observed in the normal fibroblasts. As expected, MCD-MK cells showed almost complete absence of biotin incorporation into PC, MCC and PCC (Fig. 2B). These results suggest that profound biotinidase deficiency results in a reduced level of carboxylase biotinylation in human fibroblasts in culture, even when grown in the presence of excess biotin.
Biotinidase Deficiency Hinders Recovery of Carboxylase Biotinylation in Biotin-deficient Cells in Response to Biocytin or Biotin Stimulation-The above experiment demonstrated a biotinylation defect in the MCD-BD cells that extends beyond the immediacy of the defect in releasing protein-bound biotin. To determine whether this is the result of a deficient recycling of biotin, we examined the ability of biotin-starved cells to recover carboxylase biotinylation after stimulation with biotin or biocytin. Previous studies by us and others established that nutrient biotin deficiency results in a reduction in biotinylated PC, PCC, and MCC (3,29,30). In this study, we have focused on the combined PCC-MCC biotinylation to monitor biotin status in our experiments. The cells were grown in biotin-deficient medium for 13 days and then stimulated with biotin or biocytin for 24 h at concentrations ranging from physiological to pharmacological (1, 10, and 100 nM). Total protein extracts from the cell cultures were separated by gel electrophoresis, and the recovery in PCC-MCC biotinylation was determined with streptavidin-AP. In these experiments, the results are normalized with respect to the values obtained by normal fibroblasts in normal medium. Normal fibroblasts incubated in biotin-deficient medium showed PCC-MCC biotinylation of less than 25% of the level observed in control cells. When the biotin-starved cells were stimulated with 1, 10, and 100 nM biotin, the biotin content in PCC-MCC increased to 26 Ϯ 2, 87 Ϯ 14, and 103 Ϯ 11% of control values, respectively (Fig. 3A). Similar results were obtained when biocytin, instead of biotin, was used as the supplement at the same concentrations, with the cells reaching 24 Ϯ 3, 79 Ϯ 15, and 80 Ϯ 13% of control values, respectively (Fig. 3B). These results confirm that normal cells are equally efficient in promoting carboxylase biotinylation using free biotin or biocytin as the biotin source. Next, we determined the recovery of carboxylase biotinylation in MCD-BD cells under the same experimental conditions. Biotin starvation reduced carboxylase biotin content to 37 Ϯ 5% of that observed in biotin-replete medium. Strikingly, biotinidase-deficient cells showed a poor response to biotin. When the biotin-starved cultures were stimulated with 1, 10, and 100 nM biotin, PCC-MCC biotinylation values were 33 Ϯ 6, 45 Ϯ 10, and 48 Ϯ 2%, respectively (Fig. 3C). When 1, 10, and 100 nM biocytin was added to the medium, carboxylase biotinylation was not significantly affected (27 Ϯ 7, 37 Ϯ 9, and 38 Ϯ 16%) (Fig. 3D). These results are consistent with the block in biotinidase, resulting in the inability to release biotin from biocytin for use in the biotinylation of the carboxylases. However, the reduced biotinylation in biotin-replete medium and the failure to readily respond to free biotin following biotin starvation implicates a more complex process.

Effect of Biotin Deficiency and Supplementation on Carboxylase Protein Levels in Normal and Biotinidase-deficient Cells-
We explored the relationship between the lack of response to biotin stimulation observed in MCD-BD cells and the size of the pool of apocarboxylases available for biotinylation. In this experiment we studied the effect of biotin on the recovery of carboxylase biotinylation and PC and MCC protein levels in biotin-starved normal and MCD-BD cells. As described above, Western blot analysis using streptavidin-AP revealed that biotin deficiency reduced the level of biotinylated PC, MCC and PCC in normal cells. The addition of 1 M biotin (concentra- tion 10 times higher than used above) restored carboxylase biotinylation in normal cells to control levels at 48 h. Extension of the incubation time to 72 h did not result in a significant further increase in carboxylases biotinylation (Fig. 4A). To assess the effect on carboxylase protein, Western blot analysis was performed using PC and MCC antibodies, showing that biotin deficiency in normal fibroblasts did not significantly affect the levels of these proteins. Stimulation with 1 M biotin resulted in a moderate increase in PC and MCC levels at both 48 and 72 h (Fig. 4C). In MCD-BD cells, biotin deficiency also reduced the apparent biotin content of PC, MCC, and PCC. However, unlike control cells, the reduction in carboxylase biotinylation status was matched by the level of carboxylase protein (Fig. 4, B versus D). On incubation of the biotinstarved MCD-BD in 1 M biotin, an increase in PC biotinylation could be observed at 72 h, whereas MCC biotinylation showed a significant increase at 48 h of treatment (Fig. 4B). Again, the changes in biotinylation status were matched by changes in protein levels. Although biotin deficiency reduced PC and MCC protein levels, the addition of 1 M biotin increased the amount of MCC and PC at 48 and 72 h, respectively (Fig. 4D). These results suggest that recovery of carboxylase biotinylation in normal biotin-starved fibroblasts is achieved through biotinylation of pre-existing apocarboxylases, whereas the recovery of biotinylation in MCD cells requires de novo protein synthesis.
Biotin-dependent Restoration of MCC Protein Levels Requires a Functional sGC Pathway-We had previously shown that in human cells biotin deficiency reduces the expression of most of the genes of the biotin cycle by disrupting the function of the HCS-sGC-PKG signal transduction pathway. We therefore examined the integrity of this pathway in biotin-starved normal and MCD-BD fibroblasts. First, we examined the effect of sGC inhibition on MCC expression by incubating biotin-starved normal fibroblasts in a medium containing 1 M biotin and ODQ as described under "Experimental Procedures." In these experiments, we focused on MCC protein levels because of the low PCC expression observed in MCD-BD cells. Our results show that although biotin deficiency had little effect on MCC protein levels, inhibition of sGC for 48 or 72 h results in a progressive reduction in the amount of this protein (Fig. 4E). This result suggest that in normal cells the HCS-sGC-PKG pathway is required to maintain the pool of apocarboxylases. Next, we explored the role of sGC in the recovery of MCD-BD cells.  (Fig. 4F). The results indicate that the HCS-sGC-PKG pathway is required for the restoration of carboxylase protein levels in biotin-deficient normal and MCD-BD fibroblasts. However, the poor response of MCD-BD cells to biotin suggests that expression of HCS, the key component of the pathway, may be impaired in biotinidase-deficient cells.  A and B, cell extracts were prepared from the cell cultures, and the status of carboxylases biotinylation was determined by Western blot using streptavidin-AP as described under "Experimental Procedures." C and D, the effect biotin deficiency and restoration of PC and MCC protein levels was determined by Western blot using polyclonal antibodies directed against these carboxylases. E, to explore the role of the HCS-sGC-PKG pathways on the biotin-dependent restoration of MCC protein levels, we studied the effect of inhibiting sGC on biotin-starved normal fibroblasts stimulated with 1 M biotin. Alternatively, biotin-starved MCD-BD fibroblasts were stimulated with 1 M biotin or 8-Br-cGMP, and the levels of MCC protein determined by Western blot analysis as previously described. To verify that differences in biotinylation or protein levels were not related to the amount of protein present in the samples, all of the gels were Coomassie-stained before transferring of the proteins to membranes. An example of these loading controls is depicted in A under streptavidin-AP bands. DECEMBER 5, 2008 • VOLUME 283 • NUMBER 49

JOURNAL OF BIOLOGICAL CHEMISTRY 34155
Effect of Biotin Deficiency on HCS and Biotinidase mRNA Levels in Normal and Biotinidase-deficient Fibroblasts-To test this hypothesis, we examined the impact of biotin deficiency on the level of HCS mRNA to determine whether the biotin-dependent transcriptional regulation of the biotin cycle is impaired in these cells. We used HCS mRNA as the indicator for this experiment because HCS is the key regulatory enzyme in biotin-dependent transcription and is the first enzyme of the biotin cycle subject to transcriptional control via the HCS-sGC-PKG pathway. First, we examined the time course of the effect of biotin starvation on the level of HCS mRNA, as well as biotinidase and ␤-actin mRNA, in normal cells. This was done by growing the cells in biotin-deficient medium for 13 days and assessing the mRNA levels at intervals during the course of the experiment, as described under "Experimental Procedures." The results were normalized with respect to initial mRNA levels for each culture to facilitate comparison of the results obtained with different cultures. In normal cells, biotin starvation reduced the level of HCS mRNA in a gradual manner, with 95% of the starting level remaining at day 3 and leveling off at 55-60% by days 11-15 (Fig. 5A). No significant change was observed in biotinidase or ␤-actin mRNA level during the course of the experiment. In MCD-BD cells, the level of HCS mRNA in biotin-replete medium was 52 Ϯ 3% lower than in normal fibroblasts (data not shown). Normalized to the starting level, HCS mRNA levels also fell more rapidly following the shift to biotin-deficient medium, showing a detectable reduction in the first day of the experiment and reaching 60% of the starting level by day 3 and 20% by day 15 (Fig. 5B). These results suggest that HCS mRNA levels are more easily reduced by biotin starvation in biotinidase-deficient cells than in normal fibroblasts and revealed the unexpected result that HCS mRNA was reduced even in high biotin medium. Biotinidase and ␤-actin mRNA levels, as observed for normal fibroblasts, were not affected by biotin deficiency, consistent with our previous finding that biotinidase expression is not regulated by biotin availability and the HCS-sGC-PKG pathway (3).

HCS Protein Levels and Biotinylation Activity in Biotinidase-deficient
Cells-The rapid reduction in HCS mRNA in MCD-BD cells during biotin starvation provides an explanation for the reduced level of biotinylated carboxylases in these cells compared with normal cells, because B-AMP produced by HCS is required to maintain the HCS-sGC-PKG-dependent expression of carboxylases (3,4). We therefore determined whether the reduction in HCS mRNA translated into a lower level of enzyme function. To determine HCS activity, we made use of p67, the carboxyl-terminal 67-amino acid fragment of human ␣ subunit of PCC, as the substrate for biotinylation (16,25) (Fig. 6A). Total cell extracts prepared from normal or MCD-BD fibroblasts were incubated in the presence of p67 bound to PVDF membrane and 1 Ci of [ 3 H]biotin and ATP, as described under "Experimental Procedures." The amount of membrane-fixed radioactivity, corresponding to the amount of [ 3 H]biotin incorporated in p67, was taken as a measure of HCS activity present in the different cell extracts. In this experiment, extracts from biotin-replete cells were used to minimize the presence of apocarboxylases in the samples, which would otherwise compete for substrate (5,29). Extracts from normal cells catalyzed the incorporation of 5000 Ϯ 500 cpm of [ 3 H]biotin in p67, whereas MCD-BD cell extracts showed 50% reduction in biotinylation activity, incorporating 2320 Ϯ 300 cpm into the p67 substrate (Fig. 6B). By Western blot analysis using HCS antibody, we confirmed that the reduction in biotinylation activity is associated with a reduction in the amount in HCS protein. Biotinidase-deficient cells showed one-third of the HCS immunoreactive material expressed by normal cells (Fig. 6C).

DISCUSSION
In this work, we investigated the impact of biotinidase deficiency on the maintenance of biotinylation and expression of carboxylases and regulation of the biotin cycle in human cells under conditions of biotin starvation and replenishment. Our results show that biotin deficiency, as expected, reduces carboxylase biotinylation in both normal and biotinidase-deficient cells. However, whereas in normal fibroblasts this effect is efficiently reversed after incubation with physiological concentrations of biotin or biocytin, in biotinidase-deficient cells it is not. These cells are unable to recover carboxylase biotinylation levels with biocytin and showed only partial recovery after a prolonged incubation with pharmacological doses of free biotin. The lack of response to biocytin is understood because biotinidase is required to release biotin from this compound, but the reduced response to free biotin demonstrates a more complex impact of biotinidase deficiency than is anticipated from a simple functional defect in the enzyme. Here we showed that MCD-BD behavior is associated with dimin- MCD-BD cells were expected to express normal levels of HCS, because this enzyme is not directly affected by biotinidase mutations (31). Instead, we observed a reduced level of HCS mRNA and protein and a reduced level of HCS enzyme activity in MCD-BD cells, even in high biotin medium. Further, MCD-BD cells were found to recover poorly on return to biotin-rich medium, despite a seemingly intact ability to utilize free biotin. We propose that biotinidase deficiency produces a cascade of events that begins with a block in the utilization of protein-bound biotin. The reduced access to this biotin source mimics a partial, intracellular biotin deficiency which in turn reduces the availability of B-AMP required to fuel the HCS-sGC-PKG pathway sufficiently to maintain the optimal expression of HCS and apocarboxylase genes, resulting in continued down-regulation of the pathway. A similar outcome in wild type cells, illustrated by reduction in MCC protein, could be obtained by ODQ inhibition of sGC, and full restoration of MCC protein in MCD-BD cells could be obtained on incubation of cells in 8-Br-cGMP. We surmise that once de novo synthesis of carboxylases is halted, the apparent reduction in carboxylase protein is likely the result of intracellular protein turnover. Because in normal cells, the MCC protein level was not affected by biotin deficiency, we suggest that MCC expression is maintained through endogenous biotin recycling via biotinidase activity. On the other hand, MCD-BD cells, unable to recycle protein-bound biotin, are prone to exhibit this decrease in carboxylase and HCS protein levels by their inability to fuel the HCS-sGC-PKG pathway during biotin deficiency. In MCD-BD cells, the need for a pharmacological concentration of biotin might be related to the low expression of HCS, which in turn is unable to synthesize enough biotin-BMP to activate the transcription of genes involved in the biotin cycle. However, the ability of MCD-BD cells to increase MCC protein levels when stimulated by normal concentration of 8-Br-cGMP revealed that once the block in biotin processing is bypassed, the integrity of the rest of the HCS-sGC-PKG pathway is intact. This is confirmed by the observation that 1 M biotin is unable to stimulate the increase in MCC levels in biotinstarved MCD-BD cells in the presence of ODQ.
The outcome of the effect of biotinidase deficiency on the regulation of the HCS-sGC-PKG pathway is reduced expression of HCS, PC, PCC, ACC, and sodium-dependent multivitamin transporter (3)(4)(5), producing a net deficiency of biotinylated carboxylases below the threshold required for effective contribution to intermediary metabolism. This disruption in the regulation of the biotin cycle provides an explanation for the compromised clinical state of patients with biotinidase deficiency. Although most dietary biotin is proteinbound (2,14), a simple supplementation of equivalent amounts of free biotin would have been expected to be sufficient to treat the disease. Instead, as with HCS deficiency, persistent treatment with pharmacologic doses of biotin is required to keep patients free of metabolic symptoms (2). Our studies suggest that the role of biotin supplementation is not simply to provide the product of the defective biotinidase, but it is also to overcome the repression of HCS and apocarboxylase expression that accompanies dietary biotin deficiency or genetic defects of HCS.
We are led to propose, therefore, that the constitutive expression of biotinidase may contribute to cell survival by becoming, during periods of insufficient free biotin, the primary source of biotin to fuel carboxylase biotinylation. Biotinidase is the only enzyme of the biotin cycle that is not under the transcriptional control of the HCS-sGC-PKG pathway and thus endogenous biotin recycling is expected to continue independently of biotin availability in the cell. However, in biotinidase deficiency, the brain remains as susceptible as other tissues to biotin insufficiency, resulting in a constellation of neurological abnormalities, including hearing loss, optic atrophy, loss of vision, basal ganglia calcifications, cerebral edema, cerebral atrophy, and subacute necrotizing encephalopathy (Leigh syndrome) (32, 33), that are not normally observed either in nutri- FIGURE 6. Holocarboxylase synthetase activity and protein in MCD-BD cells. HCS activity was determined in a solid phase assay in which p67 bound to PVDF membrane was used as substrate for biotinylation. A, preparation of p67 substrate. A PVDF membrane was prepared by transfer from a polyacrylamide gel following electrophoresis of extracts of E. coli XL-1 (wild type) or C-124 cells (biotin auxotroph) expressing p67. The two protein bands correspond to BCCP (upper band, 18 kDa) and p67 (lower band, 14 kDa). The membrane was developed with streptavidin-AP (left two lanes) to detect biotinylated proteins or anti-FLAG antibody (right lane) to detect unbiotinylated p67. In XL-1 cells both bands are biotinylated because of endogenously synthesized biotin. In C-124 cells, in which p67 expression was induced in biotin-free medium, only the upper band is biotinylated. The lane on the right confirms the presence of unbiotinylated p67 in the C-124 cells using the FLAG antibody. B, HCS activity of fibroblast cultures. PVDF membrane-bound p67 was incubated in the presence of [ 3 H]biotin, ATP, and total protein extracts from human fibroblasts. The amount of [ 3 H]biotin incorporated in p67 by normal fibroblasts (white bar) or MCD-BD cells (black bar) was determined using a liquid scintillation counter. Differences between normal and MCD-BD cells shown to be statistically significant (p Ͻ 0.05). C, detection of HCS protein in fibroblast cultures. The level of HCS protein in normal fibroblasts and MCD-BD cells was visualized by Western blot using an antibody against the amino-terminal region of HCS. ent biotin deficiency or in HCS deficiency. Although biotinidase mRNA and protein are expressed throughout the brain, they are most abundant in centers of the auditory and visual activity, including dorsal and ventral cochlear nuclei, superior olivary complex, and vestibular nucleus (34). Because these areas also seem to contain more biotin than other regions of the brain, it is possible that regulation of the biotin cycle may be of particular relevance for these cells (26,35). We suggest that biotinidase deficiency in combination with limited amounts of biotin may lead to the development of a secondary HCS deficiency that may compromise the survival of cells with a higher metabolic demand such as neurons. The loss of the privileged status of the brain in biotinidase deficiency is supported by necropsy data, showing an almost nonexistent PCC activity in the brain, whereas in kidney and liver, this enzyme activity is only moderately reduced (29 and 42%, respectively), with respect to the levels observed in control individual (33).
Recently, other biotin-responsive disorders have been described, including biotin-responsive basal ganglia disease and biotin-responsive limb weakness (36 -39). These diseases are characterized by subacute encephalopathy, mental retardation, severe cogwheel rigidity, dystonia, and quadriparesis. Because symptoms can be prevented or reversed with pharmacological doses of biotin and because some cases of biotinidase deficiency have been associated with progressive encephalopathy (40,41), it has been suggested that the biotin pathway is involved in these disorders. Yet the relationship to abnormalities in biotinidase, HCS, or even biotin transport has been elusive (36,37,42). It is possible that these disorders represent different degrees of the same disease in which the delicate balance between utilization of exogenous biotin and the recycling of the endogenous vitamin have been disrupted, affecting the expression of HCS and the biotin cycle.