Eur. J . Biochein. /52. 547-555 (1985) I FEBS 1985

Mammalian fatty acid synthetase is a structurally and functionally symmetrical dimer Stuart SMITH ’. Alan STERN ’ - Zafal- I . RANDIiAWA ’ and Jcns KNUDSEN ’ Research Laboratory, Children’s Hospital Medical Center, Oakland, Calilornia ’ Institulc or Biocheinislry, Odense University

(Rcccivcd March. 2h;July I S , 1985) - EJB 85 0319

We have explored a comprehensive experimental approach to determine whether the two condensingen7yme active centers of the mammalian fatty acid synthetase are siinullaneously functional. Our stratcgy involved utilization of trypsinized fatty acid synthetase, which is a nicked hoinodimcr composed of two pairs of 125 + 95- k Da polypeptides. These core polypeptides lack the chain-terminating thiocsterase domains but retain all other functional domains of the native enzyme and can assemble long-chain acyl moieties at a rate equal to that of the native enzyme. The 4’-phosphopantetheine content of these enzyme preparations, estimated from the amount of /I-alanine prescnt, from the amount of taurine formed by performic acid oxidation and from the amount of carboxymethylcysteamine formed by alkylation with iod0[2-’~C]acetate, was typically 0.86 mol/mol 95-kDa polypeptide. The stoichiometry of long-chain acyl-enzyme synthesis, measured with radiolabeled precursors, indicated that 0.84 inol acyl-chains were assembled/mol 95-kDa polypeptide. When the small ainount of apoenzyme present is taken into account, this stoichiometry translates lo 1.94 acyl chains per holocnzyme dimer. The 125-kDa polypeptide of one subunit could be cross-linked to the 95-kDa polypeptide of the other subunit by 1,3-dibromo-2-propanone yielding a single molecular species of 220 kDa. Cross-linking was accompanied by ii loss o f cnndensing-enzyme activity. This result is consistent with a structurally symmetrical model for the animal fatty acid synthetase [J. K. Stoops and S. J . Wakil (1981) J . Riol. Clzem. 256, 5128-51331 in which thc juxtaposed 4’-phosphopantetheine and cysteine thiols of opposing subunits that form the two potential catalytic centers for condensing activity are readily susceptible to cross-linking. Both half-maximal cross-linking and 50% inhi bition of activity were observed with 1 in01 1,3-dibroino-2-propanone bound/mol enzyme. After assembly of long-chain acyl moieties on the 4’-phosphopantetheine residues, no vacant condensing-enzyme active sites were demonstrable either by cross-linking with 1,3-dibroino-2-propanone or by formation of carboxymethylcysteamine on treatment with iodoacetate.

These results are consistent with a structurally and functionally symmetrical model for the mammalian fatty acid synthetase in which the two condensation sites arc simultaneously active.

In aniinals. the seven enzymes involved i n de i iow fatty acid synthesis from malonyl-CoA are integrated into two polyfunctional polypeptides. Evidence has accumulated which indicates that the fatty acid synthetase is a homodimer, each of the two polypeptides containing a single copy of the component domains of the complex [l -81. Nevertheless, synthetase monomers cannot catalyze the overall reaction of fatty acid synthesis [c), 101. This requirement for the dimeric form of the enzyme was explained through the cross-linking studies of Stoops and Wakil [I I]. They showed that the active center for the condensation reaction is forined by juxtaposi- tion of thc 4’-phosphopantetheine thiol of one subunit with an essential cysteine tliiol of the companion subunit, so that the two subunits are arranged head-to-tail. As a corollary to their head-to-tail hypothcsis. Stoops and Wakil predicted that the synthetase dimer should contain two simultaneously active centers for the condensation reaction. However, attempts to provide unequivocal proof for this hypothesis, by measuring the stoichiometry of fatty acyl-enzyme assembly, have thus

(’o,rc,.,ponr/c.r~c.r to S. Smi t h , R cscarch Laboratory, Chi Id rcn’s Hospital Medical Center, 747 52nd Street, Oakland, California. USA 94609

Ahhrcviatiori. HI’LC, high-performance liquid chromatography.

-~

far been unsuccessful. A major problem is that often the 4 - phosphopantetheine content of the enzyme is reported to be less than stoichionietric and the modified fatty acid synthetases used to measure long chain acyl-enzyme stoichiometry exhibit diminished capacity for acyl chain assembly when compared with the native enzyme [12]. I n our study we have attempted to design a careful approach which takes into account and even circumvents these difficulties.

MATEKIALS AND METHODS

EHZJ’MI<T trnd u.s*sr1J,s

Fatty acid synthetase was purilied from lactating rat niammai-y gland [13] or rat liver [14]. The fatty acid syn- thetases derived from liver and mammary gland have been used interchangeably in these studies; no differences in properties have been noticed. The trypsinized fatty acid synthetase core was isolated by proteolytic removal of the two thioesterasc domains [15]. Thioesterase I1 was purified from lactating rat maminary gland and assayed spectrophoto- metrically [I 51. Fatty acid synthetase and ketoreductase activities were measured spectrophotometrically [16]. Con-

548

densing enzyme activity was determined radiochemically as follows. Reaction mixtures, containing 0.2 M potassium phosphate buffcr (pH 6.6), 100 pM CoA, 50mM S- acetoacetyl-N-acetylcysteamine, 20 mM NaH 14C0, (20 pCi) and 42 pg cnzyrne in a final volume of 100 pl, were incubated at 37 C for 10 min before being quenched with 25 pl o f 6 % (viv) HC104. Protein was removed by centrifugation and a portion of the supernatant was spotted und dried on ;i piece of filter paper. Radioactivity on the paper was determined by liquid scintillation spectrometry.

I .3-~ibroiiio-2-propcr,zont.

Dibromopropanone (mp 27.9 C) was purified from ii

crude commercial product (Eastman) as described by Stoops and Wakil [l I]. The 1,3-dibr0ino-2-[2-'~C]propanone was synthesized from [I-'"Clbrotnoacetic acid (New England Nu- clear) ;is described by Hussain and Lowe [17]. The specific radioactivity of the product, 1.79 Ci/mol, was confirmed by determining the absorbance ( c ~ ~ ) , , , ~ , = 85 M cni I ) and radioactivity of ii solution in benzene. Fresh solutions of dibromopropanone were made for each experiment by evap- oration of the benzene and redissolution in water. The concentration of the unlabeled reagent was verified by titra- tion with dithiothreitol and estimation of the consumed thiol groups with 5,5'-dithiobis(2-nitrobenzoate) [I 81. The concen- tration of the labeled reagent was determined by liquid scintil- lation spectrometry.

C ' c i r - h o . ~ ~ ~ ~ n i c t l z ~ l ~ ~ . ~ . s t t ~ ~ i ~ i ~ i ~ i t ~

This compound was synthesized frorn mercaptoacetic acid and bronioethylamine hydrobromide essentially as described by Lindlcy [19]. The product was purified by chromatography on Dowex 50WX8 and crystallization from ethanol. The melting point of our material (168 'C) was considerably higher than that published previously ( 3 50°C) prompting us to confirm the authenticity of the product by chemical analysis. Elemental analysis, performed by Chemical Analytical Services at tlie [Jniversity of California, Berkeley, gave an empirical formula of C4 o s , Hq.os, Nl .oo , Sl .oo (oxygen not determined), agreeing well with the theoretical formula C,Hq N SOL.

A ili!~lution of : firttji ucid synthetuse tcitli ior*'oucrtute

C'arboxymethylation reactions were performed on tryp- sinized fatty acid synthetase, both untreated enzyme and en- zyme bearing long-chain acyl moieties. The procedure de- scribed helow represents the optimal conditions established for carboxymethylation with iodo[2-'"C]acetate with mini- mum loss of ["C]acyI-enzyme thioestcr.

All reagents were saturated with NZ prior to use. Long- chain acyl-enzymc was formed by incubating the trypsinized fatty acid synthetase (0.5 mg/ml) in 0.1 M potassium phos- phate buffer (pH 6.6), 10 pM acetyl-CoA, 40 pM [2-'4C]- malonyl-CoA (21.9 Ci/mol) and 100 pM NADPH for 2 min at 37 'C. Portions of the reaction mixture were removed for determination of the stoichiometry of long-chain acyl-enzyme synthesis (see below). The remaining sample (3.0 ml) was applied to a column (21 x 1.1 cm) of Sephadex G-50 equilibrated and eluted, at 4"C, with 0.05 M potassium phosphate buffer (pH 6.6), 1 mM EDTA, 0.5 mM dithio- thrcitol. 0.1 mM CoASH. This procedure removed unreacted malonyl and acetyl moieties from the enzyme by transfer to

the CoA acceptor [20] and served to separate the long-chain acyl-enzyme froin the residual substrates. In control experi- ments, enzyme was initially incubated with acetyl-CoA and [2-'4C]malonyl-CoA in the absence of NADPH and then chromatographed on Sephadex (3-50; this enzyme contained no radioactivity, thus confirming the effectiveness of the ChA treatment in unloading unrcacted substrates from the enzyme. The solution containing the major part of the protein zone was concentrated sixfold using a Centricon-10 device (Amicon) and then applied to a column (20 x 1 .I cm) of Sephadex (3-50 equilibrated and eluted at 4 ' C with a solution o f 8 M urea, 50 mM Tris, 1 mM EDTA, 0.5 mM dithiothreitol adjusted to pH 6 with HCI. The entire procedure for isolation of the long-chain acyl-enzyme took about 1.5 h. Prior and subsequent to c;~rboxymethylation, portions of the solution were removed for verification of the stability of the acyl- enzyme thioester. N o radioactivity was lost from the enzyme.

For carboxymethylation. 0.5-ml portions of the enzyme (0.5 nmol) were added to a tube containing 30 pmol sodium iodo[2-14C]acetate (4.1 6 Ciimol) and the pH was carefully acijusted to pH 8.0 by addition of 1 mM Tris base. The tube was closed under nitrogen and incubated at 37 C for 30 min. The reaction mixture was then transferred i n t o 5 ml of 10'i/o (w/v) tricliloroacetic acid at 0°C. 30 min later, tlie protein precipitate was collected by centrifugation and washed. at O'C, with 1 ml of 50OA (w/v) ethanol. The tube was drained, 0.5 ml o f 6 M HCI, containing0.09% (w/v) phenol, was added and the tube was sealed under vacuum in preparation for hydrolysis and amino acid analysis.

Elrution of :firtt.y m i d sjwtlzrtnse with I ,3-~libur~ni0-2-[2-~ 4C]proprxnone

Trypsinized fatty acid synthetase was first equilibrated with 0.1 M sodium phosphate buffer (pH 7.0), 1 mM EDTA by gel filtration on a column (4 x 1 cm) of Bio-Gel PI 50 (Bio- Rad Laboratories). Typically, trypsinized fatty acid synthetase (5 pM) was treated with 1,3-dibro1no-2-[2-'~C]prop~inone (1.79 Ci/mol, 0-4 inol/inol enzyme) for 10 min a t 70 C. Dithiothreitol ( 3 0 mM, final concentration) was added and portions of the solution were taken Cor determination of bound radioactivity and extent of cross-linking. For determi- nation of radioactivity, enzyme was precipitated, at 0 C, with trichloroacetic acid (10"/0, vjv, final concentration), collected on a Millipore filter and cd for radioactivity by liquid scintillation spectrometry. assay of cross-linking, enzyme was denatured b'y treatment with a solution containing (final concentrations) 4 M urea, 6% (v/v) niercaptoethanol. 1.25% sodium dodecyl sulfate. Portions containing I1 pg enzyme were electrophoresed on gels prepared froin 5 % (w)'v) acrylatnide 0.1 5% (w/v), bisacrylamide [21]. The procedures for gel staining and densitometric measurements have been described elsewhere [4].

In some experiments, unlabeled dibromopropanone was used, and the trypsinized fatty acid synthetase carried covalently linked long-chain ['4C]acyl moieties. I n these experiments, the gels were sliced and the radioactivity located 141.

Puotrin liyiluolysis und urnino mid analysis

Protein solutions were lyophilized in glass tubes, 6 M HCI was added and the tubes sealed under vacuum and heated for 22 h at 110°C. For taurine analysis, hydrolyzed samples were treated with 0.1 nil performic acid at 0°C for 4 h , lyophilized

549

and analyzed. Controls, consisting of either malonyl-CoA, which had been extensively purified by high-performance liquid chromatography [22] or CoASH (99% pure, Li salt, PL- Biochemicals) carboxymethylated with iodo[l -'4C]acetate, or ;I standard mixture of amino acids (Beckman), were treated similarly to cstimate recoveries of taurine, /Manine, carboxy- iiiethylcysteaniiine and other amino acids through the alkyl- ation, hydrolysis and performic acid oxidation procedures. Analyses of amino acid content, including taurine, /Manine and carboxymcthylcystearnine were performed using ;I

column (12 x 0.4 cm) of cation-exchange resin (DCSA. Durrum) eluted successively with Na 'Hi-Phi' buffer A, pH 3.25 (Dionex) for 20 min and buffer B, pH 4.25, for 30 min. The column temperature was maintained a t 50' C for 30 niin then raised to 69' C for the remaining 20 min. The column effluent, pumped at 15 ml/h, was mixed with a solution of 0.1 YO (w/v) o-phthalaldehyde, 0.2% (viv) 2-mercaptoethanol, 0.1% (vjv) Brij, 0.4 M potassium borate buffer (pH 10.4) delivered at a rate of 12 ml/h. The fluorcscent derivatives formcd were detected with a fluorometer [23]; detector rc- sponse was linear over the sample range employed. I n the analysis of [2-'"C]carboxymethylated protein hydrolysates the effluent from the fluorometer was collected in a fraction collcctor and radioactivity determined by liquid scintillation spectrometry. The presence of fluorescent material did not interfere with radioactivity determination under our con- ditions.

Asscij~ of'radiouctivitj, msociutcd with long-chin i7cj-I rlioieties co valen/i.y cittachcd to ti-lpsin i x d , f ii t ty ticid svn t he ruse

Quadruplicate portions of radioactivcly labeled long- chain acyl-enzyme (124 pg) were precipitated at 0°C with ethanol, 60% (v/v) final concentration. The protein pre- cipitate was redissolved in 0.15 ml 80'4 formic acid and 0.15 ml performic acid was added. 'The sample was left at 20 C for 2 h before 100 pg of carrier fatty acid was added. Fatty acids were then extracted by partitioning into petroleum ether ( 3 x 1 ml). Radioactivity was assayed on a portion of the extract by liquid scintillation spectrometry and the remaining fatty acid was methylated [24] for analysis on an automated gas-liquid radiochroi7iatogr~tph [25].

Samples which contained ['4C]acyl-enzyme but no resid- ual labeled substrate were also assayed for radioactivity by liquid scintillation spectrometry either directly or after first prccipitating onto Millipore filters (type RA, HA or BS) with 10% (w/v) trichloroacetic acid at 0"C. The agreement between the three methods for determination of radioactive acyl moictics was better than 6%.

Fructionution qfacyl-CoA thioesters by HPLC

We utilized an HPLC procedure, essentially as described by Corkey et al. [22], both to purify the substrates acetyl- CoA and malonyl-CoA and to analyze the CoA-ester reaction products synthesized by the trypsinized fatty acid synthetase.

Protcin cwicrnirution

Protein concentrations of fatty acid synthetase solutions were estimated from ,,m. The relationship between A 2 8 0 nm

and protein contcnt was determined by quantitative amino acid analysis of the hydrolyzed enzyme; 1 mg/ml enzyme gave A z s u ,,,,, = 1.02 i 0.01 cm- ' (n = 4).

L,iquid .s~,in/iliotioti spcctrometrj~

Scintillation fluid consisted of 33% 2-ethoxyethanol/67'%0 toluene (v/v) containing 5 mg/ml Omnifluor. Efficiency of counting was routinely determined by the channels ratio method. Efficiency of counting of radioactively labeled enzyme that had been deposited on Millipore filters was deter- mined by precipitating, on a filter, a known amount of radio- aclive protein.

Stcit isticd prrsen tat ion

Results of multiple analyses are presented as mean standnrd deviation with the number of determinations in parenthesis. The absence of a standard deviation implies a single determination.

RESULTS

Vtrlitlcit ion (? f : f iitic tional integrity f ' tryp.c.inizcJei.fi t t j > ucitl .s!nthc tmc

Our experimental approach is based on the use of the fatty acid synthetase core generated by limited tryptic digestion o f the native enzyme. We had previously shown that the core enzyme is composed of two pairs of 125 + 95-kDa poly- peptides; the 4'-phosphopantetheine moiety is associated with the 95-kDa polypeptide [26]. The core retains all the partial activities or the native enzyme except the chain-terminating thioesterase [16,26] and is potentially useful for determination of the stoichiometry or acyl-chain assembly. To validate the use of the core for this purpose, we deemed it necessary first to demonstrate that this modified enzyme was capable of assembling fatty acyl chains at the sanic rate as achieved by thc native enzyme. Two experimental approaches were used. (a) We measured directly the rate of acyl-chain assembly by the native and trypsinized enzyme (Tablc 1 A). The method requires the use of a high protein concentration to obtain adequate AA3",] "",, since the trypsinized enzyme synthesizes only stoichiometric amounts of product. The low temperature was used to facilitate more accurate measurement of the rapid NADPH oxidation rate that is inevitable with high concentra- tions of enzyme. Under these conditions we were able to measure initial rates (0-5 s) which were proportional to enzyme concentration. (b) We measured the rate of fatty acid synthesis by the core enzyme under conditions where the chain-terminating step was no longer rate-limiting (Table 1 B). This w a s achieved by addition of purified thioesterase 11, an enzyme capable of releasing acyl chains from the 4'- phosphopantetheine moiety of the synthetase [26]. Thioesterase I1 becamc non-rate-limiting above 140 pg/ml in the presence of the trypsinized enzyme and had no effect o n the rate of synthesis by the native enzyme. I n both assays there was no significant difference in the activities of trypsinized and native enzymes. Thus this modified form of the enzyme provides a suitable system for studies on the stoichiometry of acyl-enzyme formation.

Est i imt ion of tho 4'-phosphopriiz tetheirie con [en 1 q f ' trj psin i zed j i t t y acid syn thetcisr ~ n d the stoichionzetrj (? f 'tic yl-cwzyine syntlwsi,s

The 4'-phosphopantetheine content or the trypsinized fatty acid synthetase was estimated from the amount of fl-alanine and taurine formed by hydrolysis and performic acid oxidation and from the amount of [2-'4C]carboxymethyl-

Table 1 . Coinpcirison of r ( i f ~ . ~ of long-chin ucyl usscvnhly by nutivr und trypsinirrd fntty ucid synthetnse Trypsinired fatty acid synthetase was prepared from native rat livcr enzyme (spccitic activity: 2670 nmol NADPH oxidircd inin ~ ’ mg- ’ a t pH 6.6, 37 C). ( A ) Native o r trypsinized enzyme (0.1 mg, 0.2 mg) was incubated in 0.5 nil of 0.1 M potassium phosphate buffer (pH 6.6) containing 50 pM acetyl-CoA. 50 p M nialonyl-CoA and 100 yM NADPH at 10’C. (B) Native or trypsinized enzyme ( I 0 pg) was incubated in 0.5 in1 of 0.1 M potassium phosphate buffer (pH 7.0) containing 50 pM acctyl-CoA, 100 pM malonyl-CbA, 200 pM NADPI f ;inn non-ratc- limiting amounts oftliioesterase 11 jabovc 70 pg) at 3 7 ‘ C Activities were measured spcctrophotometrically as NADPH oxidi/ed a t ( A ) 10 C. pH 6.6 or (B) 37-C, pH 7.0

Conditions Initial ratcs of fatty acyl synthesis

native enzyme trypsinized enzync

nmol min ~ ’ mg-

~ ~~~ ~~ ~~~

A . Direct measurement, acyl-enzyme synthesis 327 f 24 (n = 4) 347 * 36 ( I 1 = 4)

1 8 6 0 i 2 1 ( 1 1 = 4) R. Indirect measurement, fatty acid synthesis, non-rate-limiting

thiucsterase 11 present 1890 f 97 ( n = 4)

Table 2. Antrlysis of c:j.stcine cind 4’-~/io.s~)/zo/)~~zii~l/ieini, content .f trL~sini,-ctl,futr? m i d .synt~’ietasr Values arc bascd o n a molecular mass of440 kDa. Oxidized samples had been treated with performic acid. Alkylated samplcs had been trcatcd with iodt)[’--’4C:jacctate; the carboxymethylatcd residua were assayed both by reaction with u-phthalaldehyde and by liquid scintillation spcctroinetry. The amount of cnzymc hydrolysate injectcd into the analyzer was estimated from the recovery or thc stable amino acids Acp. C~ly, Ile. Leu and Phe

Residue Trypsinized enzyme Acyl-S-trypsinized enzyme ~~ ~ ~ ~~

unmodi- oxidbed alkylated unmodi- oxidized alkylated fied fied ~~~~ ~~~~ ~~~~

rrom mass from from mass from radioactivity radioactivity

mol/tnol enzyme

Carboxymcthyl- 0 0

Csyteic acid 0.17 77.6

Carboxymethyl- 0 0

cystcine

cystcatnine

Ta LI rinc 0.03 I .96 /j-Alanine 1.99 2.04

79.8 * 1.1 (11 = 3 ) 0 (n = 3) 1.72 & 0.04 (n = 3 )

0 ( 1 1 = 3) 2.16 f 0.03 ( n = 3 )

~

~- ___________~

81.5 2 0 0 0 7 3 6 k 1 6 7 1 2 k 1 7 ( n = 3) (n = 2) ( n = 3)

~ 0 14 4 O(n = 2) -

1 72 5 0 04 0 0 0 (n = 2 ) 0 ( w = 2 ) (n = 3 ) - 0 1 86 0 (n = 2 ) -

- 2 25 1 92 1 9 0 ~ 0 0 1 -

(. = 2)

cysteamine forincd by alkylation aiid hydrolysis of the enzymc. The values for p-alanine and taurine, determined fluorimetrically as o-phthalaldehyde reaction products, usu- ally agreed closely and indicated a 4’-phophopantetheine content of 2 mol/inol enzyme (Table 2). Sometimes slightly higher values for [Manine were obtained. We attribute this to thc occasional incomplete separation of the [Manine zone from ii zone containing a minor, unidentified component. The values for carboxymethylcysteamine, determined either fiuorimetricaliy or radiochemically on the iodoacetate-treated enzyme were slightly lowei- (1.7 molimol enzyme). A typical chromatographic separation of the [2-I4C]carboxy- methylated, hydrolyzed enzyme is shown in Fig. I . A second batch of trypsinized enzyme derived from a different native enzyme preparation yielded similar results: 1.70 f 0.05 (n = 4) carboxymethylcysteamine, 1.89 0.23 (n = 3 ) taurine, 2.05 f 0.14 (n = 11) [kilanine (all ino1/4400~0 g enzyme). We established that the conditions for alkylation gave maximum yields of carboxymethylcysteamine. Thus alkylation at higher pH. for longer times, in the presence of higher concentrations of iodoacetate did not increase the yield of product; neither

did preincubatioii with dithiothreitol or sodium borohydride. Furthermorc the value for carboxymethylcysteine was close to that for cysteic acid (Table 2) . The possibility that some of the 4’-phosphopantetheine thiol was blocked by the presence of substrates (e.g. acetyl and malonyl moieties) or products (e.g. medium- or long-chain acyl moieties) was also investi- gated. Passage of the enzyme through a column of Sephadex (3-50 equilibrated with a buffer containing 0.1 mM CoA was used to allow a n y residual substrate to be transferred to the CoA acceptor. E’reincubation of the trypsinized fatty acid synthetase with thioesterase I1 (0.5 inol thiocstcrase’mol trypsinized fatty acid synthetase) was performed to remove any acyl moieties from the 4’-phosphopantetheinc. Neither of thcse procedures increased thc yield of carboxymethylcyste- amine in the subsequent reduction-alkylation reaction. These results indicate that i t is unlikely that any of the 4’- phosphopantetheine moieties of the purified enzyme arc blocked by acyl moieties and that conditions adopted for reduction and carboxymethylation of the enzyme thiols were optimal. In view of the limitations inherent i n the p-alanine analysis aiid taking into account the fact that the value for

55 1

carboxymethylcysteamine can be determined both fluori- metrically and radiochcmically, we regard the carboxymethyl- cysteamine analysis as the most reliable index of the 4'- pliosphopantetheine content of the enzyme.

Sing11 et al. [I21 have used the 'initial burst' of NADPH oxidation that accompanies assembly of the long-chain acyl moiety on the avian fatty acid synthetase as a quantitative measure of the stoichiometry ofacyl-chain synthesis. However wc havc found this approach to overestimate the real value

I I m s

d 0 10 20 30 40 5

Retention time (min) Fig. 1 . IIPLC anulysis of'[2-' C]curho~limc.tlililatrd ti-ypsinizidfutty t i c ~ i d .~twt/ie/asc~. (A) Fluorescence recording obtained with a standard mixture containing 0.95 nmol of each component. CM = carboxy- tncthyl. (B) Fluorcsccncc recording obtained with 0.18 nmol [2-I4C]- carboxymethylated trypsinized fatty acid synthetase hydrolysate. The insert shows a 10-fold amplified fluoi omctcr rcsponsc rccordcd during cmcrgcnce of carboxyniethylcysteamine and fl-alanine from the column. (C) Radioactivity profilc obtaincd simultaneously with fluo- rcscence recording shown in B. Note the change in scale on the vertical ax is

for the iiiammalian enzyme since some of the NADPH oxida- tion is associated with the synthesis of butyryl moieties which arc subsequently released from the enzyme by transfer to a CoA acceptor. It is essential, for accurate assessment of the overall stoichiometry, that only the NADPH oxidation associ- ated with the assembly of acyl moieties which reach long- chain and remain on the enzyme is taken into account. A complete inventory of the products formed by the trypsinized fatty acid synthetase from both radiolabeled malonyl-CoA and acetyl-CoA is presented in Table 3. Of the 32.6 mol NADPH oxidized/mol enzyme. approximately 2 mol was utilized in the synthesis of butyryl-CoA and 1.3 rnol for the synthesis of long-chain acyl moieties which were subsequently released as free fatty acids. The incorporation of [2-14C]- malonyl-CoA and [I -14C]acetyl-CoA into radiolabeled long- chain acyl-enzyme thioesters accounted for the oxidation of 29.0 mol NADPH. Gas chromatographic analysis revealed that the average chain length of the enzyme-linked acyl moicties was 19.M carbon atoms (Fig. 2). The estiinations for the stoichiometry based on the separate incorporations of malonyl and acetyl moieties were in complete agreement. Thus 1.63 mol of long-chain acyl thioesters was assembled/mol dimer. A separate experiment with another batch of trypsin- ized fatty acid synthetase that contained 1.7 mol4'-phospho- punktheinc/mol enzyme yielded similar results: 1.67 0.04 (11 = 4) mol long-chain acyl moieties was assembled/mol

I ' I

1 ' I

0 20 40 60 80 time (rninl

Fig. 2. Gus-liquid ratliocliroiricrtoRrrrl,hl' of 'C-lnhcltvl , f i i r /y ~icids tic~r.iivc/,frnt)i [' ' C ] r r c ~ ~ / - r ~ r i ~ , r . m ~ . Analysis was performed on a 1.8-m colutnn of 3% SE-30 on 80- 100 Supclcoport. The temperature was 1x0 C. the flow ratc 23 nil/min and the efficiency of the gas pro- portional countcr was 52%. Thc distribution of radioactivity, C j =

i 0.5% ( n = 3) was uscd to calculate the molar pcrccntagc of each acyl tnoicty synthesized. From these values, the average chain Icngth of thc products (SLY) was calculatcd using the formula: C;," = Yloo I i i ( C x niol'!h) 19.88 f 0.05 (where C = carbon nurnbcr in mol%,. 11 = number o f fatty acids in mixture). This value was uscd to calculate the number ofacyl rnoietics fornicd on the eiuyine: mol acyl inoictics synthcsized = mol malonate incorporated -8.94. A value of 1.67 k 0.04(n = 4) long-chain acyl nioietiesinmol cn7ymc was obtained

0.5k0.3'Yn, C,, = 8.1+2.4'%, C20 = 86.6?2.4%. Crr = 4.8

Table 3. Stoiciziomttry of'ptoduct fbrtnutiott by trypsinizrd, futty ticid syiitiietusc~ Results are mean k SD of three determinations. The free fatty acid had an average chain length of 16.9 i 0.1 carhon atoms; the acyl-S- enzyme had an average chain length of 19.88 k 0.05 carbon atoms

Substrate ~~~

NADP Substrate converted t o Long-chain produced ~~~~ acyl-S-enlyine fommed ~~

C4-CoA frcc fatty acid acyl-S-enzyme

mol/mol cnzymc tnol/mol inol/mol dimer holosubunit ~~~~

[2-' 'C]Malonyl-CoA 32.6 k 2.4 1.1 kO.08 0.58,0.33 14.5 k 0.71 1.63 0.95 [l -I 4C]Acetyl-CoA 32.6 k 2.4 0.85 f 0.08 0.10 k 0.03 1.63 kO.11 1.63 0.95

5 52

Fig. 3 . Tlie c?/f&ct o/' t/ibromopropLint,rlr trecitrnent on the, tnolcwikir niuss of' thc,futry ucid syntheruse COYC peprides. (A) Trypsinizcd fatty acid synthetase treated with dibromopropanone (2 mol/mol enzyme) for 5 min at 0 C; (€3) untreated enzyme. Each lane contained 10 pg enzyme. Numbers are molecular mass in kDa

enzyme, i.e. 0.98 mol/mol holosubunit. When the long-chain acyl-enzyme was subjected to carboxymethylation with iodo[2-'"C]acetate no carboxymethylcysteamine was detected either fluorimetrically or radiochemically (chromatogram not shown). This result indicated that no free 4'-phos- phopantetheine thiols were available for alkylation on the acyl-enzyme (Table 2).

Tlic usc c!f'clihroniopvopaiionc a s un artive-.vite-dire(.trd reagent

Previous work by Stoops and Wakil [ I l l had indicated that dibromopropanone was relatively specific in its action on the avian fatty acid synthetase, causing cross-linking of the two thiols at the condensing-enzyme active site. We initially performed experiments to determine whether this reagent exhibited similar specificity in its action on the native mammalian fatty acid synthctasc. The following results were obtained. (a) Fatly acid synthetase activity was completely inhibited by preincubation with 2 ~ 3 mol reagent/mol synthetase dimer. (b) Inhibition of synthetase activity by dibromopropanone was reduced when acetyl-CoA, but not malonyl-CoA, was included in the preincubation. (c) Dibroinopropanone cross-linked the mammalian synthetase subunits and maximum cross-linking was observed in thc presence of 2 - 3 mol reagent/mol synthetase dimer. (d) Cross- linking of thc synthetase by dibromopropanone was reduced when either acetyl-CoA or malonyl-CoA was included in the preincubation. (e) Thioesterase and ketoreductase activities of the synthetase wcre unaffected by dibromopropanonc. In summary, these preliminary experimental data were in complete agreement with thc carlier results of Stoops and Wakil, obtained with the avian enzyme, and indicated that

I 2 rnd dibrornopropanone bound per rnol enzyme

0

Fig. 4. Tlir c ; f ) ' i ~ t .f'[2-' C]dihvoniopropnr~c,wr. binding o i i struc.furc> iintl

firriction of'rk,f i irry acid sjntlirrrisr c o w polyppti(lc,s. The maximum observed cross-linking with this liver enzyme preparation was 80%. thc maxitnum acyl elongation was 13.1 mol malonate incorporated: mol enzyme and the maximum condensing-enzyme activity was 2.0 nmol Ht4C03 incorporatcd min- ' (mg protein)- I . Ketoreduc- tase activity (nmol NADPH oxidized min- ' mg protein-') was unaffected by dibromopropanone (not shown in figurc)

the reagent dibromopropanone causes intersubunit cross- linking of the native mammalian synthetase by linking the 4'- phosp1iopantethf:ine and cystcine thiols at condensing-en- zyrne active centers.

Dibromopropanone was then utilized as a condensing- enzyme active-sile-directed reagent for the trypsinized fatty acid synthetase. The 4'-phosphopantetheinc content of this trypsinized enzyme preparation, 1.76 0.16 mollmol ( n = 4), was identical with that of the native enzyme, 1.71 0.15 moll in01 ( P I = 4), used in its preparation.

When the trypsinized ratty acid synthetase core was treated with dibromopropanone, the amount of 125-kDa and 95-kDa species present was drastically reduced and a single new species of approximately 220 kDa was in evidence (Fig. 3). Clearly, dibromopropanone had cross-linked the 125- kDa polypeptides to the 95-kDa polypeptides. In Fig. 4 the effect of the binding of 1,3-dibromo-2-[2-'"C]propanone on cross-linking and the functional integrity of the condensing- enzyme active site is compared. The integrity of the active site was determined by measurement of the condensing-cnzynie partial activity and by measurement of the ability of the core polypeptides to incorporate malonyl moieties into ;I long- chain acyl-enzyme species. The results show that binding of approximately 1 mol dibromopropanone/mol enzyme dimer (i.e. per 440000 g of trypsinized fatty acid synthetase) is suffi- cient to achievc half-maximal cross-linking and results in a 50% loss of condensing enzyme activity and a 50% reduction in acyl-enzyme synthesis. N o effect on ketoreductase activity was observed. The maximurn extent of cross-linking observed with this enzyme preparation was 80%, consistent with the measured content of 0.85 inol 4'-phosphopantetheine/niol enzyme subunit (i.e. per mol of 125-kDa + 95-kDa species).

We next attempted to utilize dibromopropanone as a pro- be to determine whether both of the condensing-enzyme active sites can function simultaneously in the assembly of long- chain acyl moieties. The approach was first to allow the trypsinized core 1.0 synthesize the maximum possible number of long-chain acyl moieties, using radiolabeled substrate, then remove unreacted substrates from the enzyme and finally to determine whether the acyl-enzyme could be cross-linked by dibromopropanone. Polyacrylamidc gel electrophoresis of the

7. 0.501 ~

0.25 ! 24ill 1-- I 1

Gel Slice Number

Fig. 5. P ~ l . ~ u c ~ , ~ . l r r r r ~ i ~ ( ~ - g r l rl~~ctrophorc~sis of ~lihromopr,ropunonr- triwt(>d,fiii/j. ocid ~~yntlietuse core peprid(J.r. (A) Control, dibromopro- panonc-treated trypsinizcd fatty acid synthetase; (€3) dibromopro- panone-treated [14C]acyl-labeled trypsinizcd fatty acid synthetase. Molecular mass markers were intact fatty acid synthetase subunits (240 kDa) and trypsinized fatty acid synthetase core polypcptides (125 kDa i~nd 95 kDa)

dibromopanone-treated long-chain acyl-enzyme revcaled the complete absence of cross-linking, confirming that both active centers were blocked with long-chain acyl moieties. As ex- pected, all of the radioactive long-chain acyl moiety was located on the 95-kDa spccies, which is the polypeptide carrying the 4’-phosphopantetheine moiety (Fig. 5). An iden- tical experiment was performed using unlabeled acyl-enzyme and 1,3-dibromo-2-[2-’ 4C]propanone. Binding of the radio- labeled inhibitor to the acyl-enzyme was idcntical to that observed with enzymc lacking acyl moieties and was confined to the 125-kDa polypeptide (results not shown). Thcse experi- ments indicated that the presence of acyl chains on the 4 - phosphopantetheine moieties has no effect on the primary reaction of dibromopropanone with the active-sitc cysteines associated with the 125-kDa polypeptides but blocks the sec- ondary reaction of the cystcine-linked reagent with thc 4’- phosphopantetheine thiols. A schematic representation of t he reaction of dibromopropanone with the trypsinized core and the acyl-enzyme corc is shown in Fig. 6.

DISCUSSION

Two mutually exclusive models have been proposed to dcscribe the mechanism of action of the animal fatty acid synthetase. In one model, each of thc subunits alternatively acts as a catalyst and ii coordinator [27]; in the other, the two subunits are always functionally identical and are juxtaposed in a head-to-tail orientation [ I l l . These we refer to as thc functionally asymmetrical and functionally symmetrical models, respectively. Both models are consistent with the dinieric form of the enzyme being required for activity. How- ever thc asymmetrical model predicts the presence of only one site for acyl-chain assembly per dimcr, since the subunits

cannot simultaneously fulfill the roles of catalyst and co- ordinator. The symmetrical model, on the other hand, predicts that the functionally symmetrical dimer will contain two inde- pendent and simultaneously functional sites for acyl-chain assembly. I t would appear a simple matter therefore to distinguish between the two models since thcy predict different stoichiometries. Unfortunately this approach has been fraught with difficulties. Previous attempts to measure the stoichiometry have yielded values of0.9 [26], 1.2 (281 and 1.25 [I21 acyl moieties synthesized per dimer but, as pointed out by Singh et al. [12], these results may have been influenced by the presence of appreciable quantities of apoenzyme in the preparations used or by lowered capacity of the proteolytically modified enzyme for acyl-enzyme synthesis. Since, in the case of the asymmetrical model, an aposubunit unable to function catalytically may still be ablc to function as a coordinator, stoichiometries close to one acyl-chain per dimer d o not dis- tinguish clearly between the two models, regardless of the apoenzyme content. We attempted to solve these problems by adopting several independent approaches to the testing of the two models: we sought to use enzyme preparations which consisted predominantly of holoenzyme; we took care to affirm the functional integrity of the proteolytically modified enzyme used to incasure the stoichiometry ; we utilized several different direct chemical methods and one indircct method for cluantifying the 4’-phosphopantetheine content of the en- zyme; we uscd two novel approaches to determine whether all or half of the holosubunits engage in acyl-enzyme synthesis.

Data presented in this paper provide strong evidence that the corc polypeptides of the fatty acid synthetase constitute a valid system for assessing the stoichiometry of acyl-enzyme formation. Our demonstration that all of the 4-phospho- pantetheine associated with the native enzyme is retained in the trypsinized enzyme clearly refutes the claim from Wakil’s laboratory that the conditions for proteolysis adopted by us lcad to loss of the acyl-carrier peptide region from the enzyme [29]. Indced, we have found that even under extremely severe conditions of trypsin treatment there is no loss of 4’- phosphopantethcine from the core polypeptides and no fur- ther degradation of the 95-kDa polypeptide. We had prc- viously postulated that a 110-kDa polypeptide produced as an intermediate during removal of the thioesterase domain is converted to the stable 95-kDa species without loss of the peptide region containing the 4’-phosphopantetheine IS]. We cited evidence which indicated that the 15-kDa fragment re- moved during that conversion was part of the thioesterase domain [S]. The additional evidence presented here, i.e. complete retention of the 4’-phosphopantetheine moiety in the 95-kDa species and retention of full catalytic activity in the acyl-chain assembly process, supports our earlier interpreta- tion of the protcolytic cleavage pathway for the rat fatty acid synthetase. Clearly then the trypsinized rat fatty acid synthetase is a good system for elucidation of the stoichio- mctry of acyl-enzyme synthesis.

The direct approach to this question necessitates quan- titative determination of both thc 4’-phosphopantetheine available as acyl carrier and the long-chain acyl moieties formed on the enzyme. Using the dual radiochemical and fluorimetric assay of the [2-’4C]carboxymethyl-cysteamine content of alkylated enzyme, we havc identified fatty acid synthetasc preparations which consist predominantly of holoenzyme. These preparations contain typically 0.M mol 4’-phosphopantethcine/mol 95-kDa polypeptide, and exhibit an acyl-enzyme stoichiometry of 1.67 acyl chains per dimer. Such a stoichioinetry is inconsistent with a model that allows

s54

.,,,,,,a

cys +f

H? pint CY 5

native enzyme SH HS SH

W-,.

limited trypsinization

-7 trypslnlzed core HS

cys SH SH

pdnr H?

S-CHI. ,CHz-S S-CH2/C0 oc. CHz-S

crass-linked care

pdnt

SH Acyl-S acyl-enzyme core

H? pdnt CYS

S-ACYl

&-I

o n i y one ol’ the two subunits t o participate in thc catalytic Furtlierinorc, if the presence of the sinall amount o f

apo-suhunits is taken into account (14% of total subunits) then ;I stoicliiometr!. of’ 1.94 acyl chains per- diiner is obtained. This viiliie is consistent Lvith a model that allows simultaneous participation of‘ both subunits in the catalytic process. Aciiiitional results obtained by quite illdependent approaches cupport t h i s conclusion. First, the binding of 1 mol of di hroiiioprop~inoiie, mol dimcr and the accompanying cross- linking 01’ 1i:ilf ol’ the availablc 3’-phosphopnntetlieine- cont;iining polypcpidc\ results in the inhibition of‘ only 50% ot’thc condcnsing cnzymc activity of‘ the synthetase. Secondly, all of the 3‘-plinspliop~intet heine of the long-chain acyl- cnryinr i \ unavailablc for reaction with both tl i bro i n o p 1-0 pan one and i od o[ 2 - ’ ‘ C ] x e t a t e i nd icn t i ng that a I I the a\ aiitrblc prosthetic groups have engaged in acyl-enzyme lbrmation. I t should be emphasized that the laltcr es- pcriinental approach circumvents m;inq of thc potential .si)urccs of’ c i w r in the direct nicasurcmcnt of acyl-enzyme stoichioinetry. T h u s tlic approac s independent of any knowlccigc 01‘ cnzyine molecular ni ;i nd ;I bso rp t i on coe fli - cient and docs not require chemical analysis of 4’-phosphn- p:intethcinc content. The approach involves either determina- t i o n by clectrophoresis of the fraction ofacyl-enzyme that can be cross-linkcd bq dibromopropanone or measurement by H P L C of the relative iiiiiounts of r a d i o x t i v i t p eluting in the c~ i rboxymethy lc~r s t e~ i i i i i i i~ zone, using iodo[2-*“C]acetate- trciitcd, Iiytiroiy7ed free eivynie on the one hand, and long-

chain acyl-enzyme on the other. That the results of this approach agree entii-ely with the direct measurements of stoi- chiomelry provides compelling evidence indicating that the atnirnai fatty acid synthetasc is a structurally and functionall) hyinmetrical dimer a s originally predicted by Wakil et al. [30] .

Additional evidence in support of this model has recentl> been presented. Wang et al. [31] prepared subunit variants of’ fat ty acid synthetase, modified at either the 4’-phospho- pantetheine o r the cysteine rcsidues a t thc condensing-cnzyme active site. ‘l’he r:nzytiie activity of recombinant dimers w a s consistent with tvvo fu iic t i ona 1 sites fo r ac y I-e n q i n e s y n t hcsi s . Fi i i i i 1 I y. we have reex;i i n i ned the s t c) ich io me t ry of s 11 bs t I-ii tc binding to the enzyme and found the results to be consistent $5 it11 the simultaneous binding ofacetyl and inalonyl moicties a t two separate active centers 1321.

This b t u d y L V ~ S supported b! a graiit (AM 16073) f rom the hational Institutes of I Icn l th .

R F: b E R EN CES 1 . Stoopc. J. K.. Ar.;liinian. M. J.. Chalmei-s. I . H . Jr . Joshi. \’. (’ k

Wakil. S. J. (1977) in Biooyrinic r/icrnr.s[rj, ( \ a n Tanielerr. t. I:.. ed.) vol. 1 , pp. 339~--370. Acacieniic Press. N C L ~ Yorh .

2. Hcdford. c‘. .I., Kolattukudy. i’. F,. bi Roger\. La. (197X) . A i . c / r .

3. Guq. P., L A W , :S. & t-lnrdie. G. (19?X) PER.I’/xrr. 94. 33 - 3 7 . 4. I)ilccp;in, K . Pd.. Lin , C. Y. bi Smith. S. (1978) Bioi~hcwi. d. /??.

Bio~lic/?i. B i ~ p l i ~ . , ~ . 186. 139 - 1 51.

19‘) - 206.

555

S. Smith, S. & Stern. A. (1979) Arch. Biochmi. Biopliys. 107, 379-

6. Poulose. A. J.. Poster, R . J . & Kolattukudy. P. E. (1980) J . Biol.

7. McCarlhy. A . D. & Hardie, D. G. (1983) G r r . J . B i o c h n . 130,

8. Tsukarnoto, Y., Wong, H., Mattick, J . S. & Wakil, S. J . (1983) J.

9. Buttcrworth, P. f l . W.. Yang. P. C., Bock. R. M. & Porter, I . W.

10. Smith, S. & Abraham, S. (1971) .J. B id . Chtvn. ,746, 6428-6435. 11. Sloops. 1 . K . & Wakil. S. J . (1981) J . Biol. Cliein. 256, 5128-

12. Singh. N.. Wakil. S. J . & Stoops, J. K. (1984) J . B id . C ’ i z c w . ,750.

13. Smith, S. (1981) Mt,//iod,r Eiizj.tno1. 7 / , I81 - 188. 14. I.inn, T. C. (1981) Arch. Bioc,/ic~in. Biop$rs. 2OY. 613-619. 15. Smith, S. (1981) Mctliods En:nlj,niol. 7 / , 188-200. 10. Smith. S.. Agradi. E., Libertini, L. J . & Dilecpan, K . N. (1976)

17. Hussain, S. S. & Lowe, G. (1968) Biockvn . ./. /OH, 855-859. 18. Ellman, G. L. ( 1 959) Art,/7. Biochmi. Riop/ij.s. K2, 70 - 77. 19. Lindlcy. H. (1959) Air.st. J . C’lion. /2. 296-298.

387.

C‘/><J,,j. 2.55, 1 1 313- I 1 319.

185 - 193.

Rid . C/rwi. 2.~8, 1 5 3 12 - 1 5 322.

(1 967) .J. B i ~ l . C/ict?i. 242. 3508 - 351 5.

5133.

3605 - 301 I .

Proc. Nail A c d . Sci. USA 73, 11 84- 11 88.

20. Stern. A,. Sedgwick. R. & Smith, S. (1982) J . H i d . C ’ / w i n . 257.

21. Webcr. K . SC Osborn, M. (1969) .J. B id . (* / i t i n . 244. 4406-4412. 22. Corkey, B. E., Brandt, J.. Williams. R. J . & Williamson, J . K.

23. I3ensori. J . R. & Hare. P. E. (1975) Pro( , . Nut1 A c i d . .%i. USA 72,

24. MetcalPe, L. D. & Schmitz, A. A. (1961) Atitrl. C/iun. 33. 363 -

25. Agradi, f?. & Smith, S. (1970) I n / . J . Bioc/win. 7, 467--472. 26. Libertini. L. J . & Smith. S. (1979) “fr’cii. B/ot,/irni. Biop/i,r.?. IY.??

27. Kurnar. S. (1982) .J. Tiicor. Bid . 05. 263 ~ 253. 28. Scdgwick. t3. & Smith. s. (2981) /1rc/i. Bi(J(ht’fl2. Biop/ij,.c. 208,

29. Mattick. J . S.. rsukamoio, Y.. Nickless, J . & Wakil, S. .I (1983)

30. Wakil, S. J.. Stoops. J . K . & Joshi. V. c‘. (19x3) A?/i711. f l c r .

31. Wang, Y.-S., Iim. W. & flsu. R. Y . (19x4) J. B i d . C h n . ,750.

32. Mikkclson. J.. Smith. S.. Stcrn. A. & Knudsen. 1. (1985) Bioclierii.

79Y ~~ 803.

(1981) Anal. Bioc~lietn. IIK, 30-41.

619 - 622.

364.

47 ~ 60.

365 - 37‘).

J . Biol. C/ICW. .?.iA‘, 1 5 29 I - 1 5 299.

Bioc/ic,iii. 53. 537 ~ 579.

13644.- 13647.

.J.. in the press.