Regulation of Membrane Lipid Synthesis in Escherichia coli ACCUMULATION OF FREE FATTY ACIDS OF ABNORMAL LENGTH DURING INHIBITION OF PHOSPHOLIPID SYNTHESIS*

Glycerol starvation of an Escherichia coli glycerol auxotroph results in a specific inhibition of membrane phospholipid synthesis. Mindich ((1972) J. Bacteriol. 110, 96-102) observed only a trace accumulation of free fatty acid following glycerol deprivation. We have repeated these experiments using glycerol auxotrophs which also possess a lesion in beta oxidation. This defect was introduced in order to control fatty acid degradation. In contrast to the previous results, we find free fatty acid does accumulate during glycerol starvation. Similar results were found using beta oxidation-defective (fadE-) derivatives of both gpsA and plsB glycerol auxotrophs. Upon glycerol starvation of a plsB- fadE- strain, phospholipid synthesis is 90 percent inhibited. Following a lag of 20 to 40 min, free fatty acid synthesis begins and proceeds at a rate that steadily increases until the rate of fatty acid synthesis is equal to that found in glycerol-supplemented cultures. The accumulation of free fatty acid is the result of de novo synthesis. The average chain length of the fatty acid in the unesterified fraction is abnormally long. Two 20-carbon fatty acids, cis-13-eicosenoic acid and arachidic acid, are found in this frction. Furthermore, a greatly increased level of stearic acid and a small amount of a C-22 (behenic) acid are found in the free fatty acid fraction. These data indicate that acyl transfer into phospholipid is a major determinant of phospholipid acyl moiety chain length. Other experiments have shown that the free fatty acid fraction in glycerol-starved cells is metabolically active. This fraction turns over despite the defective beta oxidation system. Restoration of glycerol to starved cells allows the incorporation of the unesterified fatty acids into phospholipid.


G. ALLEN
From the Department of Molecular Biophysics and Biochemistry, Yale University School of Medicine, New Haven, Connecticut 06510 SUMMARY Glycerol starvation of an Escherichia coli glycerol auxotroph results in a specific inhibition of membrane phospholipid synthesis.
We have repeated these experiments using glycerol auxotrophs which also possess a lesion in p oxidation.
This defect was introduced in order to control fatty acid degradation.
In contrast to the previous results, we find free fatty acid does accumulate during glycerol starvation. Similar results were found using /3 oxidation-defective (fadE-) derivatives of both gpsA and PlsB glycerol auxotrophs.
Upon glycerol starvation of a PZsB-fadE-strain, phospholipid synthesis is 90% inhibited. Following a lag of 20 to 40 min, free fatty acid synthesis begins and proceeds at a rate that steadily increases until the rate of fatty acid synthesis is equal to that found in glycerol-supplemented cultures. The accumulation of free fatty acid is the result of de nova synthesis.
The average chain length of the fatty acid in the unesterified fraction is abnormally long. Two ZO-carbon fatty acids, cis-13-eicosenoic acid and arachidic acid, are found in this fraction. Furthermore, a greatly increased level of stearic acid and a small amount of a C-22 (behenic) acid are found in the free fatty acid fraction.
These data indicate that acyl transfer into phospholipid is a major determinant of phospholipid acyl moiety chain length. Other experiments have shown that the free fatty acid fraction in glycerol-starved cells is metabolically active. This fraction turns over despite the defective p oxidation system. Restoration of glycerol to starved cells allows the incorporation of the unesterified fatty acids into phospholipid. these studies have shed relatively little light on the regulation of these pathways (2). Very little free fatty acid is found during growth even of a mutant deficient in fatty acid degradation (5). Fatty acid synthesis in E. coli therefore seems tightly coupled to phospholipid synthesis (3)(4)(5). The mechanism of this coupling is unknown.
Fatty acid synthesis is constitutive and is insensitive to repression (6, 7). Feedback inhibition therefore seems to be the most reasonable regulatory mechanism. An accumulated phospholipid precursor could inhibit an early step in fatty acid synthesis and in this manner coordinate the two pathways.  Klein et al. (5).
Strain LW3 was constructed and tested exactly as described for LWl, except that the parent was Bell's (8) strain BB20-14, which owes its glycerol (or sn-glycerol 3-phosphate) requirement to a lesion in the gpsA gene. The gpsA locus is the structural gene for the biosynthetic sn-glycerol-3.phosphate dehydrogenase of E. coli (11).
All of the strains used are defective in glycerol catabolism (gZpD-).
This lesion eliminates any carbon source "shift,-down" effects that might be caused by glycerol deprivation. Ester Formation-Either the dimethoxvnronane method (9) or the diazomethane method described previously*(9) was used. Fatty acids were released from phospholipids via saponification and extracted into ether following acidification of the hydrolgsate (9) Argentation Chromatography-Monoenoic esters were separated from one another (and from saturated esters) on thin layer plates impregnated with 20% AgN03 as previously described (15). We found that Ai3 esters (of both the C-20 and the C-22 carbon acids) were well separated from (and ran ahead) of Ai1 esters such as cisvaccenate.
This result is in contrast to the behavior of Ai3 esters predicted by Morris et al. (16)  acid accumulated under these conditions. As outlined above, however, these results could be explained by degradation of the free fatty acids as well as by an inhibition of fatt,y acid synthesis.
To test this hypothesis, we examined glycerol auxotrophs which are also defective in the fl oxidation of fatty acids. Strain LWl is a glycerol auxotroph due to a defect in sn-glycerol-S-phosphate acyltransferase, the first enzyme of phospholipid biosynthesis.
The defect in /3 oxidation (fad%) lowers the activity of this system to less than lYo of the normal activity. longer in LW3 than in LWl (Fig. 3C), and the maximal rate of free fatty acid synthesis was only about half that of LWl (Fig.  3B). Strain LW3 is deficient in sn-glycerol 3-phosphate synthesis whereas LWl requires glycerol (or sn-glycerol3-phosphate) only for phospholipid synthesis. The differences between LWl and LW3, then, could be due to the lack of sn-glycerol 3-phosphate needed for the synthesis of a necessary nonphospholipid compound (although no such compound is known) in the latter strain. For this reason, we have primarily studied strain LWl. This experiment was performed as described in Fig. 1,except twice the amount of labeled acetate was used. Symbols, normalization, and abbreviations as in Fig. 1. esters and analyzed by argentation chromatography (Fig. 4). A compound running ahead of cis-vaccenate was found in the free fatty acid fraction but was not observed in either phospholinid frartion.
The behavior of this compound in this chromaboth starved and unstarved cells were converted to methyl ~~r~~~ ~~.  A culture of strain LWl was starved for glycerol and labeled with [r%]acetate after 85 min of starvation exactly as described in Fig. 1 except that the amounts of culture and labeled acetate were increased lo-fold.
The free fatty acid fraction and the phospholipid fatty acid moieties were converted to their methyl esters with diaxomethane and chromatographed in the argentation system. The esters were identified as described Co-chromatography with authentic Ai3 monoenoic esters confirmed this suggestion.
A Ai8 acid of 20 carbon atoms could be formed by elongation of cis-vaccenate through addition of an acetate unit and hence we determined the chain length of the novel acid by reversed phase chromatography.
This fatty acid was indeed 20 carbons in length (Table I). The identity of this acid was further confirmed by cleavage of the double bond with periodatepermanganate.
The monocarboxylic acid resulting from the cleavage of this acid has the same chain length as the monocarboxylic acid derived from cis-vaccenic acid. We, therefore, believe this compound is cis-13-eicosenoic acid. Analysis of the saturated fatty acid fractions was also striking (Fig. 5). A novel fatty acid was again found in the free fatty acid fraction but not in the phospholipid fractions. This acid was the 20-carbon saturated fatty acid, arachidic acid. A small amount of a 22-carbon acid (behenic acid) was also seen in the free fatty acid fraction.
Furthermore, the free fatty acid fraction is much enriched in stearic acid when compared to either of the phospholipid fractions or to wild type strains of E. coli. These alterations result in an average chain length for the saturated acids of the free fatty acid fraction that is about 2 carbons longer than the saturated acyl groups found in the phospholipid.
The over-all fatty acid compositions of these reactions are given in free fatty acid fraction of a glycerol-supplemented culture labeled by brief [r4C]acetate label was also analyzed.
The low level of radioactivity in the fraction permitted only a qualitative analysis that indicated a fatty acid composition similar to the phospholipid fraction from the same culture. A much higher ratio of cis-vaccenic acid to palmitoleic acid than is usually observed in E. coli was found in the free fatty acid fraction and in both phospholipid fractions ( Table I). The increased ratio in the phospholipids of a glycerol-supplemented culture is also observed in strain BB26-36, the fad+ parent of LWl but is not observed in strain 8, the plsB+ parent of BB26-36.
The increased content of cis-vaccenate is therefore probably due to the pZsB lesion in strains LWl and BB26-36.2 Under the present conditions, cultures supplemented with glycerol may still be slightly starved for sn-glycerol 3-phosphate and thus accumulate longer chain lengths than normal (see above). The higher levels of free fatty acid observed in the glycerol-supplemented cultures of LWl (Fig. 1) as compared to LW3 (Fig. 3)  We then restored glycerol to the culture and simultaneously removed the [r4C]acetate. Following glycerol restoration, the label in the free fatty acid fraction decreased with an equivalent increase in the label found in phospholipid (Fig. 6). These acyl chains can therefore participate in phospholipid synthesis. It should be noted that more  I  I  1  I  I  1  1  I  IO  50  90  130  170  Minutes after glycerol readdltlon FIG. 6. Incorporation of the free fatty acid fraction into phospholipid.
A culture of LWl was grown for several generations in the standard medium supplemented with 50 pg/ml of potassium oleate (to induce b oxidation).
The cells were grown to 2.5 X lO*/ml and then deprived of glycerol and oleate (see "Experimental Procedures"). One portion of the resuspended cells was supplemented with glycerol, the remainder received ["Clacetate (2.5 rCi/ml) but no glycerol. The glycerol-deprived culture was labeled for 60 min and then the cells were filtered, washed free of acetate, and resuspended in standard medium containing glycerol. At various times, samples were removed and the amounts of free fatty acid and phospholipid quantitated as described under "Experimental Procedures." At least 95% of the free fatty acid in these cells was intracellular.
A, growth of cultures. l , glycerol-supplemented; 0, glycerol starved then glycerol added at 60 min (arrow). B, distribution of counts in lipid species following glycerol addition. 0, counts per min in free fatty acid; l , counts per min in phospholipid. efficient utilization of the free fatty acid fraction was observed upon glycerol restoration if the cells had been previously induced for /3 oxidation (by growth in oleate-containing medium) before glycerol starvation.
It seems, then, that reutilization requires high levels of acyl-CoA synthetase (the only p oxidation enzyme remaining in judE mutants). Therefore, these endogenously produced fatty acids may be incorporated by a route similar to that utilized by exogenous fatty acids (4). We ha.ve observed repeatedly that the amount of lipid synthesis in glycerol-starved cultures relative to the amount of synthesis in glycerol-supplemented cultures was dependent on the length of the labeling period. Short labeling periods (5 min or less) gave relatively more lipid synthesis than long labels (20 to 40 min). Since this effect was observed with tritiated water as well as with labeled acetate, a pool effect did not seem likely. The most reasonable interpretation of these data was that the free fatty acid accumulated during glycerol starvation turned over despite the /3 oxidation block. To test this hypothesis, we did a "pulse chase" experiment on [r4C]acetate-labeled glycerolstarved cells. As shown in Fig. 7 7. Turnover of free fatty acid fraction during glycerol starvation.
Cultures of LWl were grown in the standard medium plus glycerol.
In mid-log phase, the cultures were deprived of glycerol and starved for 65 min (Experiment 1) or 100 min (Experiment 2), then labeled with acetate (see Fig. 1 Fig. 6)). After 10 min of chase, the amount of label in free fatty acid decreased with a half-life of about 30 min, indicating that the fatty acid was being degraded or altered. Our inability to chase the acetate pool efficiently implies that these data may be a considerable underestimation of the true rate of turnover.
However, the finding that free fatty acid is lost in these cultures indicates that the dependency on labeling time can be attributed to turnover of this fatty acid. Origin and Location of Free Fatty Acid I'raction-Several lines of evidence indicate that the free fatty acid fraction is the result of direct synthesis rather than of phospholipid synthesis followed by hydrolysis.
First and most compelling, the chain length distribution and species of fatty acid found in the free fatty acid fraction differ greatly from those of the fatty acid found in the phospholipids of either glycerol-supplemented or nonsupplemented cultures of LWl.
In fact, to our knowledge, fatty acids of 20 carbons have not been reported previously in E. co&. The accumulation of more than trace amounts of stearic acid is also unprecedented.
Second, if a culture of LWl is labeled before glycerol starvation (and acetate removal) no labeled free fatty acid is found upon starvation.
Third, the relatively higher rates of acetate incorporation into free fatty acid seen as the labeling period is shortened (see discussion above) also suggest that the formation of these molecules is due to de 1;ovo synthesis. The abnormally long fatty acids observed in the free fatty acid fraction (see below under "Discussion") are not the result of chain elongation of previously existing fatty acids. The l-['"Clacetate-labeled monocarboxylic acid derived from the methyl end of the 20 carbon unsaturated acid contains about 3070 of the total radioactivity. This figure is the amount expected if the entire carbon chain is synthesized during the acetate labeling period.
The free fatty acid fraction is distributed between the cells and the medium.
A portion (10 to 40%) of this fraction remains in the medium following cell removal by centrifugation or filtration.

DISCUSSIOX
The results reported in this paper lead to several conclusions concerning the regulation of lipid synthesis in L'. coli. The accumulation of free fatty acid during glycerol starvation of either type of glycerol auxotroph strongly suggests that feedback inhibition of fatty acid synthesis (by an accumulated phospholipid precursor) is not responsible for the coupling between fatty acid and phospholipid synthesis seen in this organism. In this respect this gram-negative bacterium is similar to the grampositive bacteria studied by Mindich (4). The free fatty acid accumulated during glycerol starvation contains molecules of abnormally long chain lengths. An urlsaturated fatty acid of 20 carbons in length and saturated fatty acids of 20 and 22 carbons are found. These fatty acids had not been reported previously in IS. coli. The longer chain lengths normally found in this bacterium (stearic and cis-vaccenic acids) are also enrichctl in the free fatty acid fraction at the expense of palmitic and palmitoleic acids. The in viva fatty acid synthetase system of E. coli also tends to accumulate the longer of the normal fatty acid chains (for a review see Ref. 15). However, the synthetase system functions at only a small fraction of the whole cell rate and overproduces unsaturated fatty acids. The glycerol-starved cells synthesize fatty acid at the normal rate and produce a normal ratio of unsaturated to saturated species as free fatty acid (Figs. 1 and 2; Table I). The fidelity of our system, therefore, seems much superior to that of the fatty acid synthetase and thus allows more definitive studies of the regulation of phospholipid acyl group synthesis.
The predominance of abnormally long chain lengths in glycerol-starved cells indicates that a competition between fatty acid elongation and transfer of fatty acyl groups in sn-glycerol 3-phosphate determines t.hc chain length of the acyl groups found in the E. coli phospholipid fraction.
If acyl transfer is blocked by a lack of sn-glycerol 3-phosphate, then fatty acyl thioesters continue to be elongated and appear as free fatty acids of abnormal length. It was formerly believed that the determination of chain length in E. coli was solely a property of the fatty acid synthetase system. The belief was based on the work of Greenspan et al. (23) who showed that /3-ketoacyl-acyl carrier protein synthetase was unable to elongate palmityl and cis-vaccenyl acyl carrier protein substrates although shorter chain lengths, both saturated and unsaturated, were readily elongated.
The data presented in this paper indicate that acyltransfer is an important factor determining the exact chain length of the phospholipid acyl moieties. However, some chain length specificity must also reside in fatty acid synthesis (perhaps in the @-ketoacyl-acyl carrier protein synthetase) since the chains that accumulate during glycerol starvation are (at most) only about 30% longer than normal.
Two aspects of this study were unexpected. First, a lag before free fatty acid production began was observed consistently (Figs. 1 and 2). Glycerol starvation halted phospholipid synthesis immediately but production of free fatty acid commenced only following at least 20 min of starvation.
Preliminary experiments suggest that protein synthesis is required during this lag to allow free fatty acid synthesis to commence and that the synthesis of these acids may be under catabolite control.3 Other preliminary results suggest glycerol starvation does not cause any gross changes in the intracellular acyl thioester levels4 A second surprising result was the finding that intracellular free fatty acid turned over in spite of the p oxidation lesion carried by the strains. This turnover could be attributed to incorporation of free fatty acids into a chloroform-insoluble compound, but is more likely due to traces of 6 oxidation activity not detectable by the usual assay or to a fatty acid catabolic system separate from fl oxidation.
The recent discovery of traces of fatty alcohols in .&'. coli (24, 25) suggests a possible route for a nonoxidative catabolic pathway. However, 110 increase in the levels of long chain fatty alcohols were observed following glycerol starvation.4 In conclusion, our results show that fatty acid synthesis can proceed normally in the absence of phospholipid synthesis. We have reported previously that fatty acid and phospholipid synthesis seem to be jointly regulated in experiments involving strains diploid for an unsaturated fatty acid biosynthetic gene (3). Comparison of these results with our present data suggests that the mechanism that prevents the accumulation of free fatty acid in normal cells regulates both fatty acid and phospholipid synthesis but is probably not evoked by an accumulation of intermediates in either pathway.