Activation of human mitochondrial pantothenatekinase 2 by palmitoylcarnitineRoberta Leonardi, Charles O. Rock, Suzanne Jackowski, and Yong-Mei Zhang*

Department of Infectious Diseases, St. Jude Children’s Research Hospital, Memphis, TN 38105

Edited by Gottfried Schatz, University of Basel, Reinach, Switzerland, and approved December 3, 2006 (received for review August 31, 2006)

The human isoform 2 of pantothenate kinase (PanK2) is localizedto the mitochondria, and mutations in this protein are associatedwith a progressive neurodegenerative disorder. PanK2 inhibitionby acetyl-CoA is so stringent (IC50 < 1 �M) that it is unclear how theenzyme functions in the presence of intracellular CoA concentra-tions. Palmitoylcarnitine was discovered to be a potent activator ofPanK2 that functions to competitively antagonize acetyl-CoA in-hibition. Acetyl-CoA was a competitive inhibitor of purified PanK2with respect to ATP. The interaction between PanK2 and acetyl-CoA was stable enough that a significant proportion of the purifiedprotein was isolated as the PanK2�acetyl-CoA complex. The long-chain acylcarnitine activation of PanK2 explains how PanK2 func-tions in vivo, by providing a positive regulatory mechanism tocounteract the negative regulation of PanK2 activity by acetyl-CoA.Our results suggest that PanK2 is located in the mitochondria tosense the levels of palmitoylcarnitine and up-regulate CoA biosyn-thesis in response to an increased mitochondrial demand for thecofactor to support �-oxidation.

carnitine � coenzyme A � �-oxidation � pantothenate kinase-associatedneurodegeneration

Pantothenate kinase (PanK) catalyzes the first and rate-controlling step in the biosynthesis of CoA (for review, see

ref. 1). In humans, there are three genes that express fourisoforms of PanK. PanK1� and 1� arise from the use of alternateinitiation exons of the PANK1 gene (2, 3), and the PANK3 geneproduces a single polypeptide (4). The PANK2 gene differs fromthe others in that it encodes a protein that is targeted to themitochondria (5–7). The PanK2 protein translated from the most5� start site is sequentially cleaved at two sites by mitochondrialprocessing peptidases, generating a long-lived 48-kDa matureprotein (5). Two shorter PanK2 isoforms have been described,one of which also localizes to mitochondria (7), whereas thesecond does not because of the lack of targeting sequences (6).A common feature of all PanK proteins is that they are feedbackinhibited by CoA thioesters (1); however, the isoforms aredistinguished by their unique sensitivities to the CoA thioesterpool (2–4, 8). A fourth gene, PANK4, lacks the essential catalyticglutamate residue present in all other enzymes (9) and may notbe a functional PanK.

The PanK2 isoform is the focus of much current researchbecause mutations in the human PANK2 gene give rise toPanK-associated neurodegeneration (PKAN), an inherited au-tosomal recessive disease (10). PKAN patients constitute asubset of those diagnosed with neurodegeneration with brainiron accumulation, formerly known as Hallervorden–Spatz syn-drome (11). PKAN patients have a pathological accumulation ofiron in the basal ganglia and a combination of motor symptomsin the forms of dystonia, dysarthria, intellectual impairment, andgait disturbance (11). Early-onset patients have a rapidly pro-gressive disease, and late-onset patients have a slowly progres-sive, atypical disease, where parkinsonism is common (11, 12).Two hypotheses have been proposed to explain the clinicalproblems associated with mutations in PANK2: (i) accumulationof cysteine-containing substrates because of inhibition of CoAsynthesis, which may undergo rapid autooxidation in the pres-

ence of iron, leading to free radical generation and cell damage(10); or (ii) mitochondrial CoA deficiency leading to increasedoxidative stress from free radicals (7). Both of these hypothesesposit that the mutated PanK2 proteins are functionally defective(13). Although it is clear that a number of PANK2 mutations giverise to inactive proteins, recent biochemical analyses show thatmany of the PanK2 missense mutations encode functionalproteins that are regulated by acetyl-CoA (5, 8). Disruption ofthe Pank2 gene in mice does not produce a neurodegenerativephenotype (14), although it does result in retinal degeneration,often linked with the human disease, and azoospermia, which isassociated with reduced PanK activity in Drosophila melano-gaster (15). Clearly, there is much to be learned regarding theregulatory biochemistry and role of PanK2 in intermediarymetabolism.

One of the puzzling features of PanK2 biochemistry is itsextremely high sensitivity to inhibition by acyl-CoA (5, 8). TheIC50 for acetyl-CoA is �1 �M at 2.5 mM ATP (8), which is muchlower than the cytosolic CoA concentrations estimated at 20–140�M, and, if PanK2 resides in the mitochondrial matrix compart-ment, CoA levels are likely between 2.2 and 5 mM (16, 17). Thesedata make it hard to understand how PanK2 functions in vivobecause of the high concentration of CoA thioesters. This workdescribes the reversal of acetyl-CoA inhibition of PanK2 bypalmitoylcarnitine. This new biochemical property of PanK2provides an explanation for how PanK2 is activated in vivo andsuggests that its mitochondrial location is important for itsfunction as an acylcarnitine sensor that up-regulates CoA bio-synthesis in response to accelerated demand for mitochondrial�-oxidation.

ResultsPalmitoylcarnitine Activates PanK2. One of the conundrums ofresearch on PanK2 is that it binds acetyl-CoA, and other CoAspecies, with an apparent Kd that is far below the estimatedintracellular and intramitochondrial CoA concentrations. Ac-cordingly, PanK2 expressed and partially purified from 293Tcells was inhibited by acetyl-CoA with an apparent IC50 of �0.3�M under the conditions tested (Fig. 1A). These data suggestedthat PanK2 would be inactive in vivo and led us to search forligands that may regulate PanK2. Palmitoylcarnitine was discov-ered as an activator of PanK2 activity in 293T cell lysates (Fig.1B). The enzymatic activity of PanK2 was stimulated to thehighest level by 8 �M palmitoylcarnitine. In contrast, carnitinedid not have any effect on the PanK2 activity when included inthe reactions at the same concentration range (Fig. 1B). We next

Author contributions: R.L., C.O.R., S.J., and Y.-M.Z. designed research; R.L. and Y.-M.Z.performed research; R.L., C.O.R., S.J., and Y.-M.Z. analyzed data; and R.L., C.O.R., S.J., andY.-M.Z. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS direct submission.

Abbreviations: IMS, intermembrane space; OMM, outer mitochondrial membrane; PanK,pantothenate kinase; PKAN, pantothenate kinase-associated neurodegeneration.

*To whom correspondence should be addressed. E-mail: yongmei.zhang@stjude.org.

© 2007 by The National Academy of Sciences of the USA

1494–1499 � PNAS � January 30, 2007 � vol. 104 � no. 5 www.pnas.org�cgi�doi�10.1073�pnas.0607621104

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

021

determined whether palmitoylcarnitine was able to reverse theinhibition of PanK2 by acetyl-CoA (Fig. 1C). Palmitoylcarnitinewas capable of partially reversing the inhibition of PanK2 by 0.2�M acetyl-CoA, whereas carnitine was not. Thus, palmitoylcar-nitine was a positive regulator of PanK2 activity.

Analysis of Acetyl-CoA Regulation of the Purified PanK2. The data inFig. 1 support palmitoylcarnitine as a positive regulator ofPanK2; however, interpreting the mechanism of its activation inpartially purified lysates was complicated by the lack of controlover the concentrations of endogenous regulatory ligands thatmight be present. Therefore, the N-terminal His-tagged PanK2was expressed in Escherichia coli and purified to homogeneity byaffinity and size-exclusion chromatography (Fig. 2A). The pureprotein exhibited a specific activity of 0.2421 � 0.006 pmol/ngper min under the standard PanK assay condition. The elutionposition of PanK2 on the gel filtration column was consistentwith previous work in cell lysates that indicated it was a dimer insolution. Acetyl-CoA was a potent inhibitor of purified PanK2with an IC50 value of �60 nM (Fig. 2B). We confirmed thatlong-chain acyl-CoAs, such as palmitoyl-CoA, also stronglyinhibited PanK2 activity (5, 8). The kinetic mechanism ofacetyl-CoA inhibition was explored, and the graphical analysis ofthe data indicated acetyl-CoA inhibition was competitive withATP (Fig. 2C). This conclusion was confirmed by using a directbinding assay employing the fluorescent ATP analog, TNP-ATP.The binding of TNP-ATP to proteins results in a strong increasein fluorescence that is directly proportional to the extent ofbinding. The addition of acetyl-CoA to the TNP-ATP bindingassay displaced the ATP analog from the protein (Fig. 2D),confirming that PanK2�acetyl-CoA complex formation compro-mised ATP binding.

Palmitoylcarnitine Regulation of PanK2. Palmitoylcarnitine was ableto reverse the inhibition of purified PanK2 by acetyl-CoA (Fig.3A). In this experiment, 60 nM acetyl-CoA was used, whichresulted in �50% inhibition of PanK2 activity. However, palmi-toylcarnitine stimulated PanK2 activity to a level that was�130% of the control PanK2 activity in the absence of acetyl-CoA. Palmitoylcarnitine also reversed the inhibition of PanK2 bypalmitoyl-CoA (data not shown). Three PKAN-associated pointmutants, which are active and sensitive to acetyl-CoA inhibition(8), were purified and tested for the regulation by palmitoylcar-nitine. In the presence of 60 nM acetyl-CoA, all three mutants,PanK2[R286C], PanK2[N404I], and PanK2[T528M], were inhib-ited by 40–60%. Similar to its effect on the wild-type protein,palmitoylcarnitine stimulated the activities of the mutants to

levels higher than those in the absence of acetyl-CoA (Fig. 3A).We tested the effect of palmitic acid and found that it was noteffective in the reversal of acetyl-CoA inhibition (data notshown). A kinetic analysis of the interaction between palmitoyl-carnitine and acetyl-CoA indicated that palmitoylcarnitine wasa competitive antagonist of acetyl-CoA (Fig. 3B). These datademonstrated that the mechanism of palmitoylcarnitine activa-tion was to competitively reverse the inhibition by acetyl-CoA.Paradoxically, palmitoylcarnitine also activated purified PanK2in the absence of added acetyl-CoA (Fig. 4A). Octanoylcarnitinewas not an activator indicating that the activation event is specificfor long-chain species of acylcarnitine (Fig. 4A). Fisher et al. (18)reported that carnitine was a nonessential activator of partiallypurified PanK activity from rat heart. Carnitine did not activatepurified PanK2; however, we were able to repeat the observa-tions of Fisher et al. in cell lysates showing that carnitineactivated PanK2 at higher concentrations (Fig. 4B). Thus, weattributed the activation of PanK2 by carnitine in cell lysates tothe formation of acylcarnitine from carnitine in the presence ofthe ATP:Mg2� required for the kinase assay, and we concludedthat carnitine was not a regulator of PanK2.

The PanK2�Acetyl-CoA Complex. One interpretation of the palmi-toylcarnitine activation of PanK2 in the absence of addedacetyl-CoA in the cell lysates (Fig. 1B) was the contamination ofthe sample with endogenous acetyl-CoA, but a similar activationof purified PanK2 that had been subjected to two columnchromatography steps and dialysis (Fig. 4A) suggested that aproportion of PanK2 protein was purified as the PanK2�acetyl-CoA complex. Therefore, we analyzed our pure PanK2 prepa-rations for bound acetyl-CoA. The UV spectrum of purifiedPanK2 exhibited a shoulder at 260 nm compared with thespectrum that was calculated based on the amino acid compo-sition (Fig. 5A). Heat denaturation of PanK2 followed by theremoval of the precipitated protein resulted in a supernatantwith an absorbance maximum at 260 nm (Fig. 5A), indicating thepresence of a bound nucleotide-like molecule. Based on theextinction coefficient of acetyl-CoA and the absorbance at 260nm, we calculated that between 40% and 60% of the protein indifferent batches of purified PanK2 contained a bound CoAspecies. Mass spectrometry clearly identified acetyl-CoA as thenucleotide ligand bound to PanK2 (Fig. 5B). Mass peaks corre-sponding to doubly charged (m/z � 403.72) and singly charged(m/z � 808.11) acetyl-CoA peaks were detected in the negative-ion scan. The mass spectrum gave no indication for the presenceof nonesterified CoA or other CoA species. The bacterial CoApool consists primarily of acetyl-CoA, but there are also signif-

Fig. 1. Negative and positive regulators of PanK2 activity. (A) Acetyl-CoA inhibited PanK2 activity in 293T cell lysates with an IC50 value of 0.3 �M. (B)Palmitoylcarnitine (F) activated PanK2 in partially purified 293T cell lysates, whereas carnitine (E) did not affect PanK2 activity. (C) Palmitoylcarnitine (F) reversedthe inhibition of PanK2 activity in 293T cell lysates by 0.2 �M acetyl-CoA, whereas carnitine (E) did not.

Leonardi et al. PNAS � January 30, 2007 � vol. 104 � no. 5 � 1495

BIO

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

021

icant concentrations of nonesterified CoA, succinyl-CoA, andmalonyl-CoA (19). Thus, we concluded that PanK2 selectivelycopurified with acetyl-CoA. These data show that a proportionof the total PanK2 protein survived as the PanK2�acetyl-CoAcomplex during the multistep purification protocol underscoringthe tight binding of acetyl-CoA to PanK2. The presence of thePanK2�acetyl-CoA complex explained palmitoylcarnitine activa-tion of PanK2 in the absence of added acetyl-CoA (Fig. 4A).

DiscussionThe positive and negative regulation of PanK2 by intermediatesin mitochondrial metabolism provides a mechanistic basis forunderstanding the biological function of the enzyme (Fig. 6).The mature form of human PanK2 resides in the mitochondriawhere it phosphorylates pantothenate to 4�-phosphopantothe-nate. Because the enzymes catalyzing the subsequent reactionsin the CoA biosynthetic pathway are cytosolic, the product ofPanK2 exits the mitochondria and is converted to CoASH. Toput PanK2 in the proper physiological context with its activatorand inhibitors, one has to understand where PanK2 localizeswithin the mitochondria: the intermembrane space (IMS) or thematrix. Because of the presence of an abundant outer mitochon-drial membrane (OMM) protein named porin, the OMM ishighly permeable to small molecules with molecular masses �5kDa. Thus, the cytosolic concentrations of the CoA and carnitinespecies equilibrate with the IMS. Experimental data are notavailable to pinpoint in which mitochondrial compartmentPanK2 is located. However, we propose that PanK2 is localized

to the IMS because: (i) like known IMS proteins such ascytochrome b2, cytochrome c1, and cytochrome peroxidase,PanK2 possesses two cleavable mitochondrial targeting signals(20, 21), predicting an IMS localization; and (ii) the IMSlocalization enables ready transport of pantothenate into IMSfor PanK2 activity and product release across the OMM to thecytosol where the downstream reactions in the CoA biosyntheticpathway are located. In animal tissues, the cytosolic concentra-tions of total CoA range from 20 to 140 �M (16, 17, 22).Approximately 30% (6–40 �M) of the total CoA pool areacyl-CoAs including both short- and long-chain species. Carni-tine is more abundant than CoA. The concentration of totalcarnitine is 2 mM in liver and 4 mM in heart (23). Long-chainacylcarnitine species constitute 10% of the carnitine pool,leading to a concentration ranging from 200 to 400 �M. Thus, theIMS compartment contains more activators of PanK2 thaninhibitors. The concentrations of the effectors used in our assaysare achievable in vivo. We also tested and confirmed the abilityof palmitoylcarnitine to reverse the acetyl-CoA inhibition at ahigher acetyl-CoA concentration (1 or 10 �M).

The carnitine shuttle system transports acyl groups into themitochondrial matrix and is composed of three proteins: theOMM carnitine palmitoyltransferase I that releases acylcarni-tine into the IMS, the inner membrane acylcarnitine:carnitinetranslocase, and the inner membrane carnitine palmitoyltrans-ferase II that transfers the acyl group from carnitine to CoASHin the matrix (24). Thus, matrix CoASH is required to initiatemitochondrial fatty acid �-oxidation, and a deficiency in CoASH

Fig. 2. Acetyl-CoA is a competitive inhibitor of PanK2 with respect to ATP. (A) His-tagged PanK2 was overexpressed in E. coli and purified to homogeneity byNi-affinity and size exclusion chromatography. (Inset) An SDS gel showing the purity of the PanK2 preparation. (B) Acetyl-CoA inhibited purified PanK2 with anIC50 of 60 nM. (C) Lineweaver–Burk plots of PanK2 in the absence (■ ) or presence of 30 nM (Œ) or 60 nM (�) acetyl-CoA. The lines intercept on the y axis, indicatingthat acetyl-CoA was a competitive inhibitor of PanK2 with respect to ATP. (D) Acetyl-CoA displaced the ATP analog TNP-ATP from the PanK2 active site.Fluorescent emission spectra of 5 �M TNP-ATP only (dashed line), 5 �M TNP-ATP plus 1 �M PanK2 (dotted line), and 5 �M TNP-ATP plus 1 �M PanK2 and 0.4 �Macetyl-CoA (solid line). The fluorescence of the environmentally sensitive ATP analog increased on binding to PanK2, but the addition of acetyl-CoA displacedTNP-ATP from PanK2.

1496 � www.pnas.org�cgi�doi�10.1073�pnas.0607621104 Leonardi et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

021

would lead to the accumulation of long-chain acylcarnitine. Thepotent inhibition of PanK2 by acyl-CoAs maintains the enzymein a normally off position by preventing the binding ofATP:Mg2�; however, the presence of long-chain acylcarnitineantagonizes the inhibitory action of acyl-CoA and activatesPanK2. This regulatory effect on the rate-controlling enzyme inCoA biosynthesis boosts the cytosolic concentration of CoASH,which is actively transported into the mitochondrial matrix by theLeu5p protein, an inner membrane carrier (25). The mitochon-drial localization of PanK2 places it in an ideal subcellularlocation to sense the levels of long-chain acylcarnitine and thestatus of mitochondrial �-oxidation.

In addition to its essential role in transport described in Fig.6, carnitine is also involved in modulating the acyl-CoA:CoASHratio to release CoASH to support �-oxidation, pyruvate dehy-drogenase and �-ketoglutarate dehydrogenase activities, and thetricarboxylic acid cycle (26, 27). The cellular carnitine concen-tration can approach 3 mM, and �90% of the total carnitine islocated in the cytosol (28). Like CoA, carnitine exists as freecarnitine and short- and long-chain carnitine esters, and the ratioof free to esterified carnitine varies depending on the availabilityof oxidative substrates (26, 29). In human disorders that disruptmitochondrial �-oxidation, acylcarnitines accumulate and aresecreted in the urine (30, 31), consistent with the role of carnitinein releasing CoASH to support intermediary metabolism. Our

model for the regulation of PanK2 by palmitoylcarnitine suggeststhat the absence of this enzyme activity in PKAN patients maylead to a defect in the regulation of mitochondrial matrix CoASHand the impairment of fatty acid �-oxidation. The presence ofother PanK isoforms ensures that PKAN patients are notseverely deficient in total CoA; thus, the mitochondrial dysfunc-tion in this disease would be milder than in severe mitochondrial�-oxidation disorders, perhaps accounting for the variability ofsymptoms among PKAN patients (12). The severity of symptomsin �-oxidation disorders is highly dependent on diet, and theseriousness of the clinical symptoms of the human �-oxidationdisorders can be ameliorated, in some cases, by a high-carbohydrate, low-fat diet supplemented with carnitine (32).This diet effect suggests that a similar strategy should beconsidered in the management of PKAN disease; however, thefailure of PanK2 knockout mice to accurately recapitulate theproperties of the human disease (14) makes the experimentalverification of this idea problematical in an animal model.

Materials and MethodsTransfection of HEK 293T Cells and Preparation of Cell Lysates. Thecoding sequence for the mature form of human PanK2 protein(from residue 141 of the full-length protein) was cloned intopcDNA3.1-HisA (Invitrogen, Carlsbad, CA) to yield pKM56 (8),which was used to transfect HEK 293T cells. The cells werecultured in DMEM and 10% FCS (Atlanta Biologicals, Law-renceville, GA) after transfection with FuGENE 6 according tothe manufacturer’s recommendation (Roche, Basel, Switzer-land). Cells were collected and lysed 48 h after transfection, and

Fig. 3. Effect of palmitoylcarnitine on the inhibition of PanK2 by acetyl-CoA.(A) Palmitoylcarnitine reversed the inhibition of purified wild-type (�) ormutant ([R286C], ■ ; [N404I], F; [T528M], E) PanK2 by 60 nM acetyl-CoA. (B)PanK2 activity was assayed in the presence of different concentrations ofacetyl-CoA in the absence (F) or presence of 2 �M (E) or 4 �M (■ ) palmitoyl-carnitine. Dixon plots of the reciprocal of velocity versus the acetyl-CoAconcentration showed that the lines intercepted on the y axis, indicating thatpalmitoylcarnitine activation was competitive with respect to acetyl-CoA.

Fig. 4. Activation of PanK2 was specific for palmitoylcarnitine. (A) Palmi-toylcarnitine (■ ) activated PanK2 by using the purified PanK2 preparation,whereas octanoylcarnitine (�) did not. (B) Carnitine activated PanK2 from the293T cell lysates (■ ). In contrast, carnitine did not activate purified PanK2 (�).

Leonardi et al. PNAS � January 30, 2007 � vol. 104 � no. 5 � 1497

BIO

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

021

Western blotting was used to confirm the expression of themature form of human PanK2 in the cell lysate as described (8).

Purification of His-Tagged PanK2. The PANK2 gene fragmentsencoding the mature form (amino acids 141–570) was subclonedinto expression plasmid pET-28a (Novagen, Darmstadt, Ger-many) between the NheI and BamHI restriction sites. Theresultant plasmid pKM44 was transformed into the E. coliBL21(DE3) (Stratagene, La Jolla, CA) strain to express themature form of PanK2 with an N-terminal His-tag. Transformedcells were grown at 37°C in terrific broth in the presence of 30�g/ml kanamycin to an A600 of 2 and induced with 1 mM IPTGat 18°C for 20 h. The cells were harvested by centrifugation,resuspended in MCAC-20 (20 mM Tris�HCl, pH 8/500 mMNaCl/20 mM imidazole/10% glycerol) buffer containing 1 mMPMSF, and lysed with a Microfluidizer high-pressure fluidsprocessor. The cell-free extract was precipitated with 50%saturated ammonium sulfate, and the protein pellet was resus-pended in MCAC-20 buffer containing 50 mM KSCN beforebeing loaded onto a 5-ml HiTrap Chelating HP column. Thehis-tagged PanK2 protein was eluted by using a linear gradientof imidazole from 60 to 300 mM. The fractions containingPanK2 protein were pooled, concentrated, and further purifiedto homogeneity by a Superdex 200 column (Amersham Phar-

macia Biotech, Piscataway, NJ) equilibrated in buffer (20 mMTris�HCl, pH 7.5) containing 300 mM NaCl. Three PKANdisease-associated point mutations, [R286C], [N404I], and[T528M] (numbering was based on the full-length protein), wereintroduced into pKM44 by using the Quikchange mutagenesis kit(Stratagene) according to the manufacturer’s protocol. TheHis-tagged recombinant mutant PanK2 proteins were expressedand purified as the wild-type enzyme.

PanK Activity Assay. Briefly, the standard PanK assays (3, 33)contained 45 �M D-[1-14C]pantothenate (specific activity, 55mCi/mmol; Amersham Biosciences), 250 �M ATP (pH 7.0), 10mM MgCl2, 0.1 M Tris�HCl (pH 7.5), and the indicated amountof protein from the cell extracts containing PanK2 or the purifiedPanK2 proteins in a total volume of 40 �l. The mixture wasincubated at 37°C for 10 min. The radioactive product wasquantitated by scintillation counting as described (8). The effectof different CoA or carnitine species on the enzymatic activitywas determined by including the indicated concentrations ofCoA or carnitine species in the assay mix before the addition ofPanK2. To determine the inhibitory mechanism of acetyl-CoAon the mature form of PanK2, the Km of PanK2 for ATP in thepresence or absence of acetyl-CoA was determined in standardreaction mixtures containing 180 �M D-[1-14C]pantothenate andincreasing concentrations of both ATP and the CoA thioester, asindicated.

UV-Visible Spectra of PanK2 and Mass Spectrometry Analysis. UV-visible spectra of PanK2 [0.9 mg/ml in 20 mM Tris�HCl (pH 7.4),300 mM NaCl, 1 mM EDTA, 1 mM DTT] were recorded on anAgilent 8453 spectrophotometer before and after removal of theenzyme by heat denaturation and centrifugation. The extinctioncoefficient at 260 nm for acetyl-CoA was determined to be

Fig. 5. Identification of the PanK2-bound inhibitor by UV-visible spectros-copy and electrospray mass spectrometry (ES-MS). (A) UV-visible spectra ofPanK2 (0.9 mg/ml, solid thin line) and the bound inhibitor (solid thick line)released after heat denaturation of the protein sample and removal of theprecipitate by centrifugation. The peak at 260 nm was characteristic of anucleotide-like molecule, and the spectrum of a 25 �M acetyl-CoA standardsolution (dashed line) is shown. (B) ES-MS spectrum of the supernatant froma heat-denatured sample of PanK2. The spectrum revealed the presence ofpeaks at m/z 403.72 and 808.11 corresponding to the acetyl-CoA [M-2H]2� and[M-H]� ions, respectively.

Fig. 6. A model illustrating the positive and negative regulation of PanK2activity. Human PanK2 is proposed to localize in the IMS of mitochondria,where it phosphorylates pantothenate (Pan) to form 4�-phosphopantothe-nate (P-Pan). P-Pan translocates to the cytosol and is converted to CoA by theCoA biosynthetic pathway. Fatty acid �-oxidation in the matrix requires thecarnitine shuttle system, which converts acyl-CoA to acylcarnitine in the IMS bythe action of carnitine palmitoyltransferase I (CPTI); acylcarnitine is shuttledinto the matrix by the carnitine:acylcarnitine translocase (CT) and convertedback to acyl-CoA by carnitine palmitoyltransferase II (CPTII). Acyl-CoAs, in-cluding acetyl-CoA (Ac-CoA), potently inhibit PanK2 activity by competingwith the ATP substrate. Palmitoylcarnitine reverses the acetyl-CoA inhibitionon PanK2 by competing with acetyl-CoA for the enzyme. The existence of bothnegative and positive regulation of PanK2 activity provides a mechanism toensure ample supply of CoA for �-oxidation by controlling the activity of themitochondria-localized PanK2, a rate-limiting step in the CoA biosyntheticpathway. OMM, outer mitochondrial membrane; IMM, inner mitochondrialmembrane.

1498 � www.pnas.org�cgi�doi�10.1073�pnas.0607621104 Leonardi et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

021

11,500 M�1�cm�1 from the absorbance of standard solutions andused to estimate the amount bound to PanK2.

The supernatant (100 �l) from a heat-denatured sample ofPanK2 [1.15 mg/ml in 50 mM ammonium acetate (pH 7.0)] wasdesalted on a C-8 MicroTip Column (Harvard Apparatus) formass spectrometry (MS) analysis. The column was washed withwater, and the protein-bound small molecules were eluted withmethanol (100 �l), dried under N2 flow, and resuspended inmethanol:water (1:1 vol/vol, 10 �l). MS analysis was performedby using a Finnigan TSQ Quantum (Thermo Electron Corpo-ration, San Jose, CA) triple quadrupole mass spectrometerequipped with the Nanospray Ion Source. Samples were intro-duced via static nanoelectrospray by using EconTips (NewOjective, Woburn, MA). The instrument was operated in thenegative ion mode by using single MS (Q1) scanning. Ion sourceparameters were spray voltage 1,600 V, capillary temperature270°C, and capillary offset �35 V, and tube lens offset was setby infusion of the polytyrosine tuning and calibration solution(Thermo Electron Corporation) in electrospray mode. MS ac-quisition parameters for Q1 scanning were as follows: scan range250–1,100 m/z; scan time, 0.85 s; peak width Q1, 0.7 full width,

half maximum. Instrument control and data acquisition wereperformed with the Finnigan Xcalibur (version 1.4 SR1) soft-ware (Thermo Electron Corporation).

TNP-ATP Fluorescence Spectra. The ATP analog TNP-ATP (Mo-lecular Probes, Eugene, OR) was excited at 410 nm, and thefluorescence emission spectra were recorded at 25°C on aFluoroLog spectrofluorimeter (Horiba Jobin Yvon, Edison, NJ)equipped with a circulating water bath. Binding of TNP-ATP toPanK2 and subsequent displacement by acetyl-CoA were de-tected by the change in fluorescence of a 5 �M TNP-ATPsolution in 100 mM Tris�HCl (pH 7.5), 10 mM MgCl2, 5%glycerol, after the addition of 1 �M protein and 400 nMacetyl-CoA.

We thank Ruobing Zhou and Karen Miller for their expert technicalassistance and Phil Poston (Hartwell Center of St. Jude Children’sResearch Hospital) for the mass spectrometry experiments. This workwas supported by National Institutes of Health Grants GM 62896 (toS.J.), Cancer Center (CORE) Support Grant CA 21765, and theAmerican Lebanese Syrian Associated Charities.

1. Leonardi R, Zhang YM, Rock CO, Jackowski S (2005) Prog Lipid Res44:125–153.

2. Rock CO, Calder RB, Karim MA, Jackowski S (2000) J Biol Chem 275:1377–1383.

3. Rock CO, Park HW, Jackowski S (2003) J Bacteriol 185:3410–3415.4. Zhang YM, Rock CO, Jackowski S (2005) J Biol Chem 280:32594–32601.5. Kotzbauer PT, Truax AC, Trojanowski JQ, Lee VMY (2005) J Neurosci

25:689–698.6. Hortnagel K, Prokisch H, Meitinger T (2003) Hum Mol Genet 12:321–327.7. Johnson MA, Kuo YM, Westaway SK, Parker SM, Ching KH, Gitschier J,

Hayflick SJ (2004) Ann NY Acad Sci 1012:282–298.8. Zhang YM, Rock CO, Jackowski S (2006) J Biol Chem 281:107–114.9. Hong BS, Yun MK, Zhang YM, Chohnan S, Rock CO, White SW, Jackowski

S, Park HW, Leonardi R (2006) Structure (London) 14:1251–1261.10. Zhou B, Westaway SK, Levinson B, Johnson MA, Gitschier J, Hayflick SJ

(2001) Nat Genet 28:345–349.11. Thomas M, Hayflick SJ, Jankovic J (2004) Mov Disord 19:36–42.12. Hayflick SJ, Westaway SK, Levinson B, Zhou B, Johnson MA, Ching KH,

Gitschier J (2003) N Engl J Med 348:33–40.13. Hayflick SJ (2003) J Neurol Sci 207:106–107.14. Kuo YM, Duncan JL, Westaway SK, Yang H, Nune G, Xu EY, Hayflick SJ,

Gitschier J (2005) Hum Mol Genet 14:49–57.15. Afshar K, Gonczy P, DiNardo S, Wasserman SA (2001) Genetics 157:1267–1276.

16. Idell-Wenger J, Grotyohann L, Neely JR (1978) J Biol Chem 253:4310–4318.17. Williamson J, Corkey B (1979) Methods Enzymol 55:200–222.18. Fisher MN, Robishaw JD, Neely JR (1985) J Biol Chem 260:15745–15751.19. Vallari DS, Jackowski S, Rock CO (1987) J Biol Chem 262:2468–2471.20. Gasser SM, Ohashi A, Daum G, Bohni PC, Gibson J, Reid GA, Yonetani T,

Schatz G (1982) Proc Natl Acad Sci USA 79:267–271.21. Van Loon AP, Brandli AW, Pesold-Hurt B, Blank D, Schatz G (1987) EMBO

J 6:2433–2439.22. Horie S, Isobe M, Suga T (1986) J Biochem 99:1345–1352.23. Ramsay RR (1994) Essays Biochem 28:47–61.24. Kerner J, Hoppel C (2000) Biochim Biophys Acta 1486:1–17.25. Prohl C, Pelzer W, Diekert K, Kmita H, Bedekovics T, Kispal G, Lill R (2001)

Mol Cell Biol 21:1089–1097.26. Ramsay RR, Zammit VA (2004) Mol Aspects Med 25:475–493.27. Bieber LL (1988) Annu Rev Biochem 57:261–283.28. Hutter JF, Alves C, Soboll S (1990) Biochim Biophys Acta 1016:244–252.29. Ramsay RR, Arduini A (1993) Arch Biochem Biophys 302:307–314.30. Sim KG, Carpenter K, Hammond J, Christodoulou J, Wilcken B (2002)

Metabolism 51:366–371.31. Wanders RJ, Vreken P, den Boer ME, Wijburg FA, van Gennip AH, Ijlst L

(1999) J Inherit Metab Dis 22:442–487.32. Vockley J, Whiteman DA (2002) Neuromuscul Disord 12:235–246.33. Rock CO, Karim MA, Zhang YM, Jackowski S (2002) Gene 291:35–43.

Leonardi et al. PNAS � January 30, 2007 � vol. 104 � no. 5 � 1499

BIO

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

021