CEMISTRY THE JURNAL OF BIOWOGICAL

Vol. 257, No. 18, Issue of September 25, pp. 10759-10765, 1982 Printedin US.A.

Regulation of Phospholipid Synthesis in Escherichia coli COMPOSITION OF THE ACYL-ACYL CARRIER PROTEIN POOL IN VIVO* (Received for publication, April 2, 1982)

Charles O. Rock and Suzanne Jackowski From the Department of Biochemistry, St. Jude Children's Research Hospital, Memphis, Tennessee 38101

The regulation of membrane lipid biogenesis was investigated by measuring the levels of the acyl-acyl carrier protein (acyl-ACP) intermediates in the biosynthetic pathway. In particular, the role of the sn-glycerol-3-phosphate acyltransferase was assessed by focusing on the size and composition of the long chain acyl-ACP pool. The ACP pool was specifically labeled in vivo with fi-[3- 3H]alanine and the ACP subspecies analyzed by reversed phase liquid chromatography and conformationally sensitive gel electrophoresis. The acyl-ACP pool was found to be a small fraction of the total ACP in normally growing cells and was particularly devoid of chain lengths that could serve as acyltransferase substrates. Inhibition of phospholipid synthesis at the acyltransferase step resulted in a rapid increase in the content of acyl-ACP, and analysis showed the presence of chain lengths that are acyltransferase substrates. Acyl-CoAs were not detected during interruption of acyl transfer activity. These results show that 1) acyl-ACPs are the acyl donors for phospholipid synthesis in vivo, 2) the acyltransferase does not play a role in the regulation of the lipid biosynthetic rate or the composition of phospholipid acyl moieties, 3) the primary regulatory site in phospholipid biosynthesis is at an early step in fatty acid biosynthesis, 4) feedback regulation by long chain acyl-ACP's is not a controlling mechanism for fatty acid synthesis under normal physiological circumstances, and 5) enzymes that utilize acyl-ACPs are involved in kinetic competition for the scarce acyl-ACP substrates.

Although many molecular details need to be resolved, the biosynthetic pathways leading to the major lipid classes of Escherichia coli have been established (1-7). Considerably less information is available on the regulation of these pathways and their coordination with other cellular activities (58). Previous investigations have focused on two possible regulatory points or rate-controlling steps in membrane lipid formation (5, 6, 8). First, phospholipid synthesis could be regulated by controlling the production of acyl moieties, most probably at an early step in fatty acid biosynthesis. There are several possible candidates for the rate-limiting step at this stage. Two examples are regulation of malonyl-CoA production by acetyl-CoA carboxylase (4, 5) and regulation of the active ACP content by modulating a component of the pros* This work was supported by National Institutes of Health Grant GM 28035, Cancer Center (CORE) Support Grant CA 21765 from the National Cancer Institute, and American Lebanese Syrian Associated Charities. 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. 1The abbreviations used are: ACP, acyl carrier protein, includes

thetic group turnover cycle (1, 7, 8). A second possibility is that the incorporation of acyl moieties is regulated at the snglycerol-3-phosphate acyltransferase step (6-9). This enzyme is the first committed step in phospholipid biosynthesis and has also been implicated in the regulation of membrane fatty acid composition (6-9). The choice between these two possibilities is not clear from the available information, since the supporting data are primarily derived from in vitro experiments on isolated enzyme systems (7, 8). The rate-limiting step in membrane lipid biogenesis in vivo could be directly determined by measuring the relative pool size of the intermediates in the biosynthetic pathway. All of the intermediates exist as thioesters of ACP (1, 10), and the two possible regulatory scenarios suggest markedly different ACP pool compositions during balanced growth. If an early step in fatty acid biosynthesis is rate-limiting, then unesterifled ACPSH or perhaps acetyl-ACP would be the predominant intracellular form of ACP. On the other hand, if the acyltransferase reaction is rate-limiting and responsible for controlling the fatty acid composition, then a significant pool of long chain acyl-ACP substrates would be found in vivo. Thus, quantitation of the size and composition of the acylACP pool in vivo would decide between the two alternatives. In the one previous study on the intracellular acyl-ACP content (11), 20% of the ACP pool in exponentially growing cells was found to be resistant to S-alkylation unless first treated with neutral hydroxylamine. It was concluded from these data that 20% of the ACP pool was acylated (11), but the identity of the putative acyl moieties was not determined. In this report, we have used extensions of recently developed techniques for the analysis of ACP derivatives (12-17) to examine the composition of the acyl-ACP pool in greater detail. EXPERIMENTAL

PROCEDURES

Materials-Sourcesfor supplies were: New England Nuclear, #f-[3H]alanine (specific activity, 30 Ci/mmol); Whatman, DEAE-cellulose (DE52); Amersham, scintillation mixture (ACS); Analtech, thin layer plates; Altex, Perisorb RP-8 column; Bio-Rad, electrophoresis and gel filtration media and supplies. All other chemicals and solvents were reagent grade or better. ACP was purified as described previously (15, 18). [1- 14 C]Acyl-ACP standards were prepared biosynthetically and purified as detailed previously (12, 14). Short chain (2 to 6) acyl-ACP was a gift from Dr. J. Cronan (University of Illinois) and was prepared by the acyl-imidazole method (19). BacterialStrains and Growth Conditions-Allstrains were derivatives of E. coli K12. The genetic lesion, panD, results in the inactivation of aspartate 1-decarboxylase (20, 21), the only enzyme in E. coli capable of providing fi-alanine for pantothenate biosynthesis (20, 3

all species of ACP; ACPSH, acyl carrier protein known to have a free sulfhydryl; acyl-ACP, acyl-acyl carrier protein; (ACPS) 2, acyl carrier protein dimer linked through the prosthetic group sulfhydryls; ACPSSG, acyl carrier protein mixed disulfide with glutathione; acyltransferase, sn-glycerol-3-phosphateacyltransferase.

10759

10760

Acyl-ACP Pools in E. coli

22). Thus, panD strains have an absolute requirement for ,8-alanine, and the ACP and CoA pools can be uniformly labeled with exogenous ,f-[3-3H]alanine (16). All strains were made auxotrophic for fl-alanine by transduction mediated by P1 phage using a tetracycline resistance determinant (zad220::Tn10)that is closely linked to the panD locus as the selected marker (16, 22). The strains utilized and their genotypes were: SJ16 (panD, zad220::Tn10,metB, relA1, AR, A-, naiR, F-) derived from UB1005 (16); SJ22 (plsB26,panD, zad220::Tn10,glpR, glpD, glpK, relA1, spoT1, tonA22, T2 R, pit-10, HfrC) derived from

BB26-36 (23); SJ62 (gpsA, panD, zad220::Tn1, glpD, glpR, glpK, relA1, spoT1, tonA22, T2R,pit-10, HfrC) derived from BB20-14 (23, 24); and SJ32, a plsB+ of SJ22 obtained by transduction with wild type phage. In all experiments, strains were grown on minimal medium E (25) containing thiamine (0.001%) and glucose (0.4%) as the carbon source. In addition, methionine (0.01%) for strain SJ16 and glycerol (0.05%) for strains SJ22 (plsB) and SJ62 (gpsA) were included to satisfy the growth requirements of these strains. For each experiment, overnight cultures of the panD strains were grown on 1 m 8-[3-3 H]alanine (specific activity, 30 Ci/mmol) to insure that the pools of [3 H]CoA and [3 H]ACP have the same specific activity as the input fi,-alanine. During the experiment, 2 yM f-[3-H]alanine (specific activity, 30 Ci/ mmol) was included in all cultures to satisfy the growth requirement for 8-alanine and to continue labeling the ACP and CoA pools uniformly. Doubling times of thepanDstrains on 2 AM fi-alanine were identical with the doubling times of their isogenic panD+ parents. Cultures were grown at 37 C, and cell number was measured using a Klett-Summerson colorimeter calibrated by determining the number of colony-forming units per ml as a function of colorimeter (klett) readings.

Depriving Cells of Glycerol or ,B-Alanine-At mid-exponential phase of growth (4 to 6 x 108 cells/m), 10 ml of the growing culture was filtered through a 0.45 u Millipore HAWP filter and washed twice with 5-ml portions of prewarmed media without glycerol or f-alanine. The filter was then placed in 10 ml of prewarmed media without the required supplement and resuspended by vigorous shaking. The filtration procedure consiuned 2-3 min, and the time of deprivation began when all of the original medium was through the filter. As was found by others employing this method (26), growth of plsB or gpsA strains continued for about one generation following removal of the glycerol from the culture. The extent of growth of SJ16 following /alanine deprivation was representative of panD strains in general and depended on the amount of ,8-alanine initially presented to the cells, since the level of the CoA reserves in strain SJ16 varies with f-alanine input (16). In this experiment, strain SJ16 was grown on 1 iM fi-[33 H]alanine to a density of 4 x 108 cells/ml and then deprived of the supplement as described above. Growth continued at the same rate for about one generation, and then the cells divided once more at onehalf the original rate. Liquid Chromatographyof ACP-Samples of total [3 H]ACP species were extracted from fi-[3- 3 H]alanine-labeled cells using the 2propanol procedure described previously (16). The 2-propanol supernatants were loaded onto a 1-ml DEAE-cellulose (DE52) column that was washed with 5 ml of 10 mM bis-Tris, pH 6.5, to remove the 2propanol and unbound components. The [3H]CoA was eluted from the column with 0.2 M LiCI in the same buffer, and [H]ACP subsequently eluted with 0.45 M LiCl in the column buffer. Radiochemical purity of these compounds was determined by thin layer chromatography (16), and the [3 H]ACP in the 0.45 M LiCI fraction was >99% pure. Aliquots of this fraction were used directly for chromatographic analysis. Acyl-ACP species were resolved according to chain length using an adaptation of the method of Rock and Garwin (12). A column (0.45 x 50 cm) packed with the 8-carbon reversed phase pellicular support, Perisorb RP-8, and a Waters liquid chromatographic system were employed. Separations were achieved using gradients of 2-propanol in 50 mM potassium phosphate, pH 7.3. Flow rate was 0.5 ml per min, and 1.0-ml (2 min) fractions were collected. A 9-37% linear gradient of 2-propanol was capable of resolving a wide range of acyl-ACPs, and under these conditions, the elution position of a particular acylACP species was generally reproducible to ±+0.5% 2-propanol. The separations were qualitatively similar to those achieved previously using octyl-Sepharose (12, 14) in that a linear relationship between the logarithm of the elution position (as defined by the percentage of 2-propanol) and carbon number was found. Monounsaturated long chain acyl-ACPs have carbon numbers that are one less than their actual chain length (12, 14). Standards for ,8-keto, f-hydroxy-, and 2trans-acyl-ACP were not available, but since the interaction of acyl-

ACP with reversed phase supports involves the terminal portion of the acyl moiety (12, 14, 19), these derivatives would most probably have carbon numbers equivalent to their chain length. (ACPS) 2 was adsorbed more tightly to Perisorb RP-8 than ACPSH and eluted in the same region as acyl-ACP. Short chain acyl-ACP, ACPSSG, and other ACP-mixed disulfides with polar thiols had elution positions identical with ACPSH. The distribution of ACP species was determined by counting a 200-,d aliquot of each fraction in 3 ml of ACS. Recovery of total radioactivity was greater than 75%, and between 0.5 and 4.0 x 106 cpm of [3H]ACP were used for each analysis.

ElectrophoreticAnalysis of ACP Species-Samples were prepared by Triton X-100 extraction of cells using the procedure of Clewell and Helinski (27). Aliquots of the Triton X-100 lysates were loaded directly into the gel slots immediately following lysis. Conformationally sensitive gel electrophoresis was performed using 20% gels as described previously (13, 15, 17), except that the Hoefer SE600 gel apparatus was used, and the temperature of the gel was maintained at 37 C during electrophoresis. Temperature was found to be a critical variable in obtaining consistent separations. If the gels were run at 15-20 C, faster migrations and almost no resolution of ACP species were observed. On the other hand, ACP migrated more slowly on gels run at 50-60 C, but no separation of species was observed, and acyl-ACP was deacylated. This electrophoresis system does not distinguish between 8:0-ACP and 16:0-ACP (13), but was useful as another method for determining the total proportion of acyl-ACP in the extract. The electrophoresis technique compliments the reversed phase chromatography in that 2:0-, 4:0-, and 6:0-ACP are resolved from each other (17), and ACP mixed disulfides, like ACPSSG, also have distinctly different mobilities from ACPSH or acyl-ACP (13, 15). The RF values of standard compounds were: (ACPS)2, 0.29; ACPSSG, 0.65; ACPSH, 0.69; 2:0-ACP, 0.73; 4:0-ACP, 0.78; 6:0-ACP, 0.81; 8:0ACP through 16:0-ACP, 0.84. When added, the dithiothreitol concentration was 10 mM. Gels (0.75 mm) were processed for fluorography by fixation in 5% trichloroacetic acid (2-16 h) followed by two 15-min water washes (200 ml) and a 15-min exposure to 1.5 M sodium salicylate (28). Gels were dried at 65 C and exposed to Kodak X-Omat AR film at -80 °C. Quantitation of fluorograms was performed using a single-photoncounting Ortec 4310 densitometer. The information was captured using a multichannel analyzer (Ortec 9220 nanosecond fluorimeter) and digital data was passed through a Hewlett-Packard 9845A computer for integration routines and graphic display. Tube gels (5-mm diameter) were also run using the same gel composition as the slab gels. Tube gels were frozen, sliced into 1-mm segments, and crushed in 0.2 ml of water. After standing overnight to elute the [3H]ACP, the

Fraction Number

FIG. 1. Size and composition of the acyl-ACP pool in exponentially growing cells. Strain SJ16 was labeled with 2 pM f-[33H]alanine and sampled in mid-logarithmic phase growth (about 6 X 108 cells/ml). A, separation of total acyl-ACP from unesterified ACP. B, fractionation of the acyl-ACP pool according to carbon number. Insets show 10 x expansion of the Y-axis and a carbon number calibration scale. Reversed phase chromatography was performed as described under "Experimental Procedures." In both cases, samples were reduced with 10 mM dithiothreitol just prior to chromatography, although unreduced samples gave virtually identical results.

10761

Acyl-ACP Pools in E. coli slices were counted in 3 ml of ACS. We found very good agreement

between these two methods of quantitating the electrophoretic separations. RESULTS

Acyl-ACP in Growing Cells-In order to determine the acyl-ACP content of E. coli during normal exponential growth, strain SJ16 was cultured on glucose minimal media containing 2 M -[3-3H]alanine, and at various intervals during growth, cell samples were extracted and the ACP pool purified and fractionated as described under "Experimental Procedures." In several experiments, reversed phase chromatography showed that the acyl-ACP content was between 8 and 12% of the total pool (Fig. 1A) in strain SJ16 sampled in the middle of logarithmic growth (6 x 108 cells/ml). This estimate of the total acyl-ACP pool was also found in all other samples of strain SJ16, regardless of the phase of growth. Fractionation of the acyl-ACP pool by gradient elution of Perisorb RP-8 showed that the acyl-ACP species present were not evenly distributed (Fig. 1B). Acyltransferase substrates (carbon number >15) were poorly represented (Fig. 1B), comprising at most a few tenths of a percent of the total. The average ACP content measured in our experiments was 8 pmol/10 cells. Setting the lower limit for detection of an acylSJ16 log phase

0.8

ACP (94%)

b 0.6 x

U 0.4 0.2

ACYL-ACP

02 0.4 0.6 0.8 Relative Mobility

FIG. 2. Acyl-ACP content of logarithmically growing cells. Conformationally sensitive gel electrophoresis was performed as de-

scribed under "Experimental Procedures" and used to fractionate the [3H]ACP pool extracted from log phase SJ16. The slice-and-count method (see under "Experimental Procedures") was used to determine the distribution of radioactivity in the gel.

ACP chain length at 0.1% of the total, it can be calculated that at most approximately 50 molecules of each acyl-ACP chain length are available to the acyltransferase at any given time in vivo. The main acyl-ACP component (Fig. 1B) appeared to have a carbon number of 14.5 and did not co-migrate with any of our acyl-ACP standards. This acyl-ACP was the predominant form in all SJ16 extracts isolated from both logarithmic and stationary phase cells. The composition of the ACP pool was also assessed by conformationally sensitive gel electrophoresis (Fig. 2). Since many more samples could be processed using gel electrophoresis, the ACP composition of strain SJ16 was analyzed every 30 min (doubling time = 60 min) throughout the growth cycle and into stationary phase. In general, quantitation of the radioactive distribution on the gels was achieved by densitometry (see under "Experimental Procedures"), but the sliceand-count method was also applied to many samples (Fig. 2). These two methods gave the same value for the acyl-ACP content. Acyl-ACP was found to comprise between 4 and 8% of the total (Fig. 2), and no differences were found between any of the samples in the growth study. The gel electrophoresis method consistently gave a lower estimate of the acylACP content as compared to the liquid chromatographic technique. One conceivable explanation for this observation is that the high pH (9.0) and temperature (37 °C) of the separating gel may hydrolyze a portion of the thioesters during the run. However, acyl-ACP is quite stable (29) and forms tight bands in the gel (13, 15, 17). If degradation was occurring, tailing of [3H]ACP back from the acyl-ACP band would also have been observed. No evidence was found for 2:0-, 4:0-, or 6:0-ACP by the electrophoresis techniques, and (ACPS) 2 was not found in freshly prepared samples. This technique also revealed the presence of an ACP species possessing a slower mobility than ACPSH and comprising approximately onethird of the total ACP pool. In the presence of reducing agent (10 mM dithiothreitol), the radioactivity in this band disappeared and was found in the ACPSH band and, therefore, was identified as ACPSSG on the basis of its change in mobility following reduction and its co-migration with authentic ACPSSG standard. ACPSSG has been identified as a major species of ACP obtained from stationary phase E. coli B cells (13).

1.6

FIG. 3. Fractionation of the CoA pool on Perisorb RP-8. The CoA fraction of strain SJ22 (plsB) starved for glycerol was loaded onto the RP-8 column that was developed with a linear gradient from 8-15% 2-propanol followed by a 70% 2-propanol wash beginning at fraction 75, all in 50 mM potassium phosphate, pH 7.3. Conditions for reversed phase chromatography and sampling are described under "Experimental Procedures." Greater than 98% of the radioactivity eluted with CoASH.

1.2

X 0O

0.4

0.0 0

10

20

30

48

FRRCTION

50

NUMBER

68

70

80

Acyl-ACP Pools in E. coli

10762

Analysis of the CoA Pool-Extensive analysis of the CoA pool was not a goal of this work, but a few observations were made relevant to the presence of long chain acyl-CoAs in vivo. All of our lysates were analyzed by thin layer chromatography (16) to determine the initial proportion of CoA and ACP in the sample. Long chain acyl-CoAs are resolved from CoA by the solvent system employed (16), and no indication of a significant acyl-CoA pool was evident from these data. The CoA fraction from the DEAE-cellulose column was also analyzed on Perisorb RP-8 under conditions similar to those used for the resolution of acyl-ACPs (Fig. 3). Like authentic CoASH, the vast majority (96%) of the CoA did not bind to the column (Fig. 3). The [1- 14 C]16:0-CoA standard, in contrast,

N

o

ro o_

u

=

o

Fraction Number

FIG. 4. Reversed phase chromatography of the ACP fraction before and after glycerol deprivation of strain SJ22. Strain SJ22 (plsB) was grown, deprived of glycerol, sampled, and the acylACP pool prepared as described under "Experimental Procedures." A, determination of the total long chain acyl-ACP content of strain SJ22 (plsB) during logarithmic phase growth (4 x 108 cells/ml). B, Long chain acyl-ACP 10 min following glycerol deprivation of strain SJ22.

bound to the column and was not eluted until the high 2propanol wash. The sample in Fig. 3 is from SJ22 (plsB) that had been starved for glycerol, but the pattern shown is similar to what was found with SJ16. Less than 0.2% of the CoA pool eluted in the position of long chain acyl-CoA (Fig. 3), and thus it appears that the levels of acyl-CoA in E. coli are very low. Acyl-ACP Pool in Cells Starved for /f-Alanine-Previous investigators (11) have reported a large accumulation of acylACP following starvation of cells for pantothenate. In our experiments, strain SJ16 was deprived of fi-alanine in the midlogarithmic state of growth (see under "Experimental Procedures"), and samples were taken before deprivation and every 30 min after removal offi-alanine. Thin layer chromatography (16) showed that the cellular CoA pool dropped precipitously following starvation for f,-alanine (see also Ref. 11). Changes in the size and composition of the acyl-ACP pool were not evident during starvation. Gel electrophoresis showed no increase in acyl-ACP content above the 5 to 8% value normally found for this strain during log phase growth. Similarly, reversed phase liquid chromatography did not show any change in the acyl-ACP pool composition, and chromatograms almost identical with that shown in Fig. 1 were obtained. Acyl-ACP Pool in the Absence of PhospholipidSynthesisTwo strains were used in these studies. Strain SJ22 (plsB) possesses an acyltransferase with a requirement for glycerolP 10 times higher than wild type (23, 30), and strain SJ62 (gpsA) has an inactive, biosynthetic glycerol-P dehydrogenase (23, 24) and thus cannot manufacture its own glycerol-P. Both strains are phenotypically recognized as glycerol (or glycerol-P) auxotrophs (23, 24), and removal of the glycerol from the growth medium results in an immediate cessation of phospholipid synthesis (23, 26), although growth continues for about one generation (23, 26), All experiments described in this section were performed on both strains, and essentially no difference was found in the behavior of these strains. The acyl-ACP pool in exponentially growing SJ22 (plsB) was the same as was found in strain SJ16 (Fig. 4A). Acyl-ACP comprised about 10% of the total pool, and fractionation of this pool gave a result similar to that shown in Fig. 1A. The acyl-ACP pool was found to increase 4-fold following removal of the glycerol from the medium (Fig. 4B). Gradient elution showed that the acyl-ACPs that accumulate following cessa-

I.O

1.0 V -- I

FIG. 5. Fractionation of the acylACP pool in the absence of phospholipid synthesis. The ACP fraction was obtained and chromatographed as described under "Experimental Procedures" from strain SJ22 (plsB) that had been deprived of glycerol for 10 min. The RP-8 column was developed with a 937% gradient of 2-propanol followed by a 65% 2-propanol bump that began at fraction 75. The inset shows the elution position of standard acyl-ACP chain lengths. Acyl-ACP comprised 44% of the total recovered radioactivity in this run.

0.7

x fi_ 0U U

0.5

0.2

0.0 0

10

20

30

40

50

FRRCTION NUMBER

60

70

80

10764

Acyl-ACP Pools in E. coli

arithmic growth (Figs. 1 and 2) suggests that an early step in the pathway governs the rate of fatty acid biosynthesis. There are four possible candidate enzymes, but acetyl-CoA carboxylase (4, 5) seems to be the most plausible choice. The ratelimiting role of this enzyme in fatty acid biosynthesis seems well established in mammalian systems, and its activity is controlled by both allosteric (38) and covalent modification (39) mechanisms. However, a mechanism for the control of this enzyme in E. coli has yet to be established. Another possible mechanism could operate by modulating the level of active ACP molecules (8). This would be achieved by modifying a component of the prosthetic group turnover cycle to change the ratio of apo- to holoprotein in the cell. Regulation at the acetyl or malonyl transacylase steps also cannot be ruled out. Clearly, more definitive evidence is needed to decide the exact site and mechanism that governs the rate of membrane lipid accumulation. The direct transfer of acyl moieties from acyl-ACP to glycerol-P is indicated as the normal de novo pathway for the incorporation of fatty acids into phospholipid. This reaction is catalyzed by the membrane-bound acyltransferase that can utilize either acyl-CoA or acyl-ACP as acyl donors in vitro (9, 40, 41). Acyl-ACP seemed to be the most plausible in vivo acyl donor, since it is well established that acyl-ACPs are the end product of fatty acid biosynthesis (42), and higher specificity in the acylation reaction in vitro was found with acylACP as compared to acyl-CoA (41). However, due to the lack of definitive in vivo information, the identity of the physiological acyl donor has been considered unresolved (9). Phospholipid synthesis can be specifically arrested at the acyltransferase step by withdrawing the glycerol supplement from either a gpsA or plsB mutant (23, 26). In our experiments, an abrupt increase in the acyl-ACP content occurred immediately following inhibition of acyltransferase activity in both these strains (Figs. 4 and 6). Moreover, the acyl-ACPs that accumulate (Fig. 5) have the same chain lengths that are found in E. coli phospholipids and that are most active in the acyltransferase assay in vitro (41). In these experiments, no evidence of an acyl-CoA pool was obtained (Fig. 3). These data directly tie the utilization of acyl-ACP in vivo to the activity of the acyltransferase and leave little doubt that acyl-ACP functions as the physiological acyl donor. Our results also argue that the rate of fatty acid biosynthesis is not coupled to the rate of phospholipid biosynthesis. This point has received considerable attention (23, 26, 43, 44), and these experiments have focused on the events following the deprivation of gpsA oyr lsB mutants of the required glycerol supplement. Phosphohpid synthesis is immediately inhibited >95% following removal of glycerol (23, 26), although growth, macromolecular synthesis, and assembly of membrane protein constitutents continues normally for about one generation before the cells cease to divide (26). Cronan et al. (43) reported that free fatty acids of abnormally long chain lengths accumulate in the absence of phospholipid synthesis. The conclusion of Cronan et al. (43) that fatty acid and phospholipid synthesis are not tightly coupled has been challenged by Nunn et al. (44), who concluded that fatty acid and phospholipid synthesis are coordinately inhibited during glycerol starvation. Our in vivo results show that immediately following removal of the glycerol supplement, the acyl-ACP content increases sharply (Fig. 4). Thus, fatty acid biosynthesis continues in the absence of phospholipid synthesis, and these two processes are not tightly coupled. However, our data also suggest that fatty acid biosynthesis should be severely reduced following glycerol starvation. As a result of the increase in acyl-ACP, ACPSH almost disappears immediately following glycerol starvation (Fig. 6) and does not recover to its prestarvation

level during the 2-h course of the experiment (Fig. 7). The rate of fatty acid biosynthesis would be depressed under these circumstances due to the lack of ACPSH that is required for new rounds of synthesis. Thus, our data provide a rationale for the apparently conflicting observations published previously (41, 42). Feedback inhibition of an early step in fatty acid biosynthesis by acyl-ACP is a possible mechanism for controlling fatty acid production, but seems to be ruled out by our results. If end product inhibition by long chain acyl-ACP is operating to maintain the low level of these intermediates found in logarithmically growing cells (Figs. 1 and 2), then an increase in acyl-ACP content above this level would not occur following the inhibition of phospholipid synthesis at the acyltransferase step. In fact, the opposite result was found. Following glycerol starvation of plsB or gpsA mutants, essentially all of the ACPSH becomes esterified with long chain acyl moieties (Figs. 5 to 7). Therefore, we conclude that feedback control by acyl-ACP is not a significant factor in modulating fatty acid biosynthesis. The concept that feedback regulation may be a mechanism for governing the production of acyl moieties arose primarily from the observation that the addition of exogenous fatty acids to the growth media resulted in a decrease in the content of acyl moieties derived from de novo biosynthesis in the membrane phospholipids (45-48). The biochemical intermediates involved in the incorporation of these exogenous fatty acids into phospholipid have not been identified, and the pathway may or may not involve entry into the acyl-ACP pool (7, 8). Despite this difficulty in interpreting these experiments, the change in membrane composition is best explained as a competition for acylation between endogenously and exogenously derived fatty acids rather than by the inhibition of de novo biosynthesis (49). One perplexing point concerns the composition of the acyl moieties that accumulate during glycerol starvation. Cronan et al. (43) found longer than normal chain lengths in the free fatty acid pool, but these chain lengths were not observed by us in the acyl-ACP pool (Fig. 5). The reason for this is unknown, but it is possible that a thioesterase exists that is more active on abnormally long chain acyl-ACP, thus functioning to remove the inappropriate chain lengths from the acyl-ACP pool. This idea is consistent with the observed disappearance of acyl-ACP with time following glycerol starvation (Fig. 7). In fact, the time course of acyl-ACP disappearance coincides almost exactly with the appearance of the free fatty acids seen in the previous experiments (41, 50). This precursor-product relationship strongly suggests that the free fatty acids are metabolically derived from acyl-ACP, and since the composition of the two pools is different, a specific thioesterase is indicated. It is possible that one or both of the soluble thioesterases (51, 52) are responsible for the observed disappearance of acyl-ACP. Although the thioesterases have a low reactivity toward acyl-ACP, their activity increases substantially with increasing chain length (53). In any event, it is doubtful that this phenomenon has a significant impact on the pathway during balanced growth, since the half-time for hydrolysis in vivo (15 min) is long compared to the division cycle. Acknowledgments-We wish to thank Dr. John Cronan for his gift of short chain acyl-ACPs and Kim Lee for his skillful technical assistance. REFERENCES 1. Vagelos, P. R. (1973) in The Enzymes (Boyer, P. D., ed) 3rd Ed, Vol. 8, pp. 155-199, Academic Press, New York 2. Silbert, D. F. (1975) Annu. Rev. Biochem. 44, 315-339 3. Cronan, J. E., Jr., and Gelman, E. P. (1975) Bacteriol.Rev. 39,

Acyl-ACP Pools in E. coli 232-256 4. Volpe, J. J., and Vagelos, P. R. (1976) Physiol.Rev. 56, 339-417 5. Bloch, K., and Vance, D. (1977) Annu. Rev. Biochem. 46, 263298 6. Raetz, C. R. H. (1978) Microbiol. Rev. 42, 614-659 7. Cronan, J. E., Jr. (1978) Annu. Rev. Biochem. 47, 163-189 8. Rock, C. O., and Cronan, J. E., Jr. (1982) Curr. Top. Membr. Transp. 17, 207-233 9. Green, P. R., Merrill, A. H., Jr., and Bell, R. M. (1981) J. Biol. Chem. 256, 11151-11159 10. Prescott, D. J., and Vagelos, P. R. (1972) Adv. Enzymol. Relat.

Areas Mol. Biol. 36, 269-311 11. Elovson, J., and Vagelos, P. R. (1975) Arch. Biochem. Biophys. 168, 490-497 12. Rock, C. 0., and Garwin, J. L. (1979) J. Biol. Chem. 254, 71237128 13. Rock, C. O., Cronan, J. E., Jr., and Armitage, . M. (1981) J. Biol. Chem. 256, 2669-2674 14. Rock, C. O., Garwin, J. L., and Cronan, J. E., Jr. (1981) Methods Enzymol. 72, 397-403 15. Rock, C. O., and Cronan, J. E., Jr. (1981) Methods Enzymol. 71, 341-351 16. Jackowski, S., and Rock, C. 0. (1981) J. Bacteriol. 148, 926932 17. Cronan, J. E., Jr. (1982) J. Biol. Chem. 257, 5013-5017 18. Rock, C. O., and Cronan, J. E., Jr. (1980) Anal. Biochem. 102, 362-364 19. Cronan, J. E., Jr., and Klages, A. L. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, 5440-5444 20. Cronan, J. E., Jr. (1980) J. Bacteriol. 141, 1291-1297 21. Williamson, J. M., and Brown, G. M. (1979) J. Biol. Chem. 254, 8074-8082 22. Cronan, J. E., Jr., Littel, K. J., and Jackowski, S. (1982) J. Bacteriol.149, 916-922 23. Bell, R. M. (1974) J. Bacteriol. 117, 1065-1076 24. Cronan, J. E., Jr., and Bell, R. M. (1974) J. Bacteriol. 118, 598605 25. Vogel, H. J., and Bonner, D. M. (1956) J. Biol. Chem. 218, 97106 26. McIntyre, T. M., and Bell, R. M. (1975) J. Biol. Chem. 250, 90539059 27. Clewell, D. B., and Helinski, D. R. (1969) Proc. Natl. Acad. Sci. U. S. A. 69, 2110-2114 28. Chamberlain, J. P. (1979) Anal. Biochem. 98, 132-135

10765

29. Rock, C. O., and Cronan, J. E., Jr. (1979) J. Biol. Chem. 254, 9778-9785 30. Bell, R. M. (1975) J. Biol. Chem. 250, 7147-7152 31. Snider, M. D. (1979) J. Biol. Chem. 254, 7197-7202 32. Silbert, D. F. (1970) Biochemistry 9, 3631-3640 33. Cronan, J. E., Jr., and Gelmann, E. P. (1973) J. Biol. Chem. 248, 1188-1195 34. Jackson, M. B., and Cronan, J. E., Jr. (1978) Biochim. Biophys. Acta 512, 472-479 35. Marr, A. G., and Ingraham, J. L. (1962) J. Bacteriol. 84, 12601267 36. Cronan, J. E., Jr. (1975) J. Biol. Chem. 250, 7074-7077 37. Garwin, J. L., Klages, A. L., and Cronan, J. E., Jr. (1980) J. Biol. Chem. 255, 3263-3265 38. Yeh, L.-A., Song, C.-S., and Kim, K.-H. (1981) J. Bial. Chem. 256, 2289-2296 39. Lent, B., and Kim, K.-H. (1982) J. Biol. Chem. 257, 1897-1901 40. Ray, T. K., and Cronan, J. E., Jr. (1975) J. Biol. Chem. 250, 84228427 41. Rock, C. O., Goelz, S. E., and Cronan, J. E., Jr. (1981) J. Biol. Chem. 256, 736-742 42. Cronan, J. E., Jr., and Vagelos, P. R. (1972) Biochim. Biophys. Acta 265, 25-60 43. Cronan, J. E., Jr., Weisberg, L. J., and Allen, R. G. (1975) J. Biol. Chem. 250, 5835-5840 44. Nunn, W. D., Kelly, D. L., and Stumfall, M. Y. (1977) J. Bacteriol. 132, 526-531 45. Silbert, D. F., Cohen, M., and Harder, M. E. (1972) J. Biol. Chem. 247, 1699-1707 46. Sinensky, M. (1971) J. Bacteriol.106, 449-455 47. Silbert, D. F., Ulbright, T. M., and Honegger, J. L. (1973) Biochemistry 12, 164-171 48. Estroumza, J., and Ailhaud, G. (1971) Biochimie (Paris)53, 837839 49. Polacco, M. L., and Cronan, J. E., Jr. (1977) J. Biol. Chem. 252, 5488-5490 50. Ray, T. K., and Cronan, J. E., Jr. (1976) Biochem. Biophys. Res. Commun. 69, 506-513 51. Barnes, E. M., Jr., and Wakil, S. J. (1968) J. Biol. Chem. 243, 2955-2962 52. Bonner, W. M., and Bloch, K. (1972) J. Biol. Chem. 247, 31233133 53. Spencer, A. K., Greenspan, A. D., and Cronan, J. E., Jr. (1978) J. Biol. Chem. 253, 5922-5926