The association of phosphatidylserine synthetase with ribosomes in extracts of Escherichia coli.

Abstract The biosynthesis of phospholipids in Escherichia coli is catalyzed by enzymes most of which are associated with the bacterial membrane. In the present work, it has been found that the enzyme CDP-diglyceride: l-serine phosphatidyltransferase (phosphatidylserine synthetase) is exceptional in that it is not bound to membrane fragments in cell-free extracts of E. coli, but is predominantly associated with ribosomes. The enzyme resists extraction from ribosomes in buffers of high ionic strength. When ribosomes were allowed to dissociate in buffers containing low concentrations of magnesium, the enzyme was associated with both 50 and 30 S subunits. Phosphatidylserine is found only in trace amounts in E. coli, being rapidly converted to phosphatidylethanolamine by a decarboxylase which is a membrane-bound enzyme. Despite the apparently different intracellular localization of the two enzymes, their function in vivo must be tightly coupled. When the decarboxylase in living cells of E. coli was inhibited by the addition of hydroxylamine to the medium, phosphatidyl[1-14C]serine was shown to accumulate within 1 min. In the control cells, in which the decarboxylase was not inhibited, no labeled phosphatidylserine could be detected, showing that the decarboxylation of the newly formed phosphatidylserine must take place without detectable time lag.


Preparation of Cell Extracts
The bacteria were harvested at the end of log phase at a concentration of log per ml. In a typical experiment 1 liter of bacterial suspension was centrifuged at 0", and the cells were washed once with 100 ml of the same buffer, usually 10 mM Tris of pH 8.0, employed subsequently for sonication. Portions of the bacterial suspension (5 to 10 ml) were disrupted at 0" with an MSE loo-watt sonicator at an amplitude of 7 to 8 urn and a frequency of 22 kc. Four 30-s bursts were sufficient to assure complete disruption, with halfminute intervals to allow cooling of the sample.
Sonicates were subsequently stored on an ice bath. Pencillin lysates were made by a modification of the procedure of Lederberg (7). Cells grown to a density of lo9 per ml on Tris-glycerol were treat.ed with the potassium salt of penicillin G (added as a concentrated solution) at a level of 1 mg per ml. After an hour of further shaking at 37" over 99% of the cells had lysed as judged by the viability count.
Freshly grown E. coli B were also lysed by the freeze-thaw-lysozyme-deoxycholate method (with 0.3% detergent) as described by Ron et al. (8).

Diflerential Centrifugation and Sucrose Gradient Analysis
These were performed at O-2". The gradients were prepared in 12-ml polyallumer tubes ( which was linear with enzyme concentration, was allowed to proceed for 10 min at 30". A unit of activity is defined as the amount of enzyme which forms 1 nmole of product per min. To assay the phosphatidylglycerophosphate synthetase (Scheme 1, Reaction 4) the incubation conditions described by Chang and Kennedy (3) were employed, except that the final reaction volume was 0.06 ml.
A unit of activity was defined as above, except that the incubations were carried out at 37". Diglyceride Kinase-This enzyme catalyzes the synthesis of phosphatidic acid from ATP and 1,2sn-diglycerides (10, 11). A final reaction volume of 0.1 ml was employed.
iZfter 10 min at 37", the reaction was stopped with 1.5 ml of chloroform-methanol, the lipid phase was extracted with 2 M KC1 solution, and 0.1 ml of the CHCls phase was counted directly in Patterson-Green fluid. Phosphatidylserine Decarbosylase-This enzyme (Scheme 1, Reaction 3) was assayed as described by Kanfer and Kennedy (2). A unit of activity of decarboxylase or diglyceride kinase was defined as for the other enzymes.
None of these enzymes were inhibited by 2 to 10% sucrose in the incubation mixture, under conditions in which the samples from gradient centrifugations were assayed. Portions of cells were suspended in 0.01 M Tris buffer of pH 8.0 (except in Experiment 6), containing various additions as indicated, sonically disrupted, and subjected to differential centrifugation at 0". The synthetase activities of the unfractionated sonicates so prepared were closely similar, and each is taken as 100% in calculating the recoveries in the supernatants as shown. The activity sedimented at 100,000 X g for 5 hours was fully recovered in the pellet fraction.
Loosely packed material above the pellet was decanted and assayed with the supernatant fraction. After the position of the lipids had been determined by staining with iodine vapor, l-cm strips were cut out and counted.
Nearly all of the radioactivity was recovered in a sharp band coincident with marker authentic phosphatidylserine.

Other Procedures
Protein was determined by the method of Lowry et al. (21). RNA was measured by the pentose reaction with orcinol (13) with yeast RNA as a standard.
The presence of deoxycholate in the samples caused a slight turbidity during the orcinol reaction; this was removed by centrifugation without interference with the RNA determination.

Sedimentation
of Phosphatidylserine Xynthetase during Differential Centrijugation-When extracts prepared by sonic disruption of cells of E. coli B were subjected to differential centrifugation, the pattern of sedimentation of phosphabidylserine synthetase strongly suggested association of this enzyme with ribosomes (Table  I).
Thus, when centrifugation was carried out in a medium of low magnesium concent,ration (1 mM), little or no activity was sedimented during centrifugation at 40,000 X g for 1 hour.
The fraction precipitated under these conditions contains membrane fragments, and is enriched in membranebound enzymes.
When the 40,000 x g supernatant fraction was subjected to centrifugation at 100,000 X g under conditions expected to sediment ribosomes, the greater part of the enzyme activity was recovered in the pellet. (Table I). Furthermore, increasing the magnesium concentration to 10 mM, which is known to cause aggregation of ribosomes, caused the sedimentation of 43$'& of the enzyme act.ivity in the 40,000 X g pellet', together with a comparable percentage of the RNA of the estract.
Essentially similar results were obtained upon analysis of cells lysed by the gentler penicillin method, indicating that the result obtained is not an artifact produced by sonic disruption of the cells.
The sedimentation pattern was identical when cells were grown on minimal medium 63 (5) with glycerol as carbon source, or when minimal glucose was employed. Extracts prepared from frozen cells of E. coli B also presented much the same pattern.

Treatment of Ribosomal
Pellets with Bu$ers qf High Ionic Strength-When ribosomal pellets collected by centrifugation at 100,000 i: g were treated with buffers containing high concentrations of salt, or 50 InM EDTA, the bulk of the enzyme activity remained associated with the ribosomal pellet upon subsequent recentrifugation (Table II)  L-a-glycerophosphate phosphatidyltransferase (PGP synthetase).
The distribution of these enzymes in a sucrose gradient was measured under the same conditions as Fig. 1, and compared to phosphatidylserine decarboxylase as a marker. Large membrane fragments were removed by a preliminary centrifugation at 30,000 X g for 30 min. A sample (0.4 ml) was applied to a 5 to 20y0 sucrose gradient made up in the same buffer.
After centrifugation at 200,000 X g for 2 hours at 2", samples were analyzed for phosphatidylserine synthetase and optical density at 260 nm as indicated.
these conditions (2 mM MgClJ most of the ribosomes are undissociated; the second, smaller peak of synthetase activity is probably coincident with dissociated ribosomes. In an effort to localize the synthetase on the subunits of the ribosomes, the experiment shown in Fig. 3  Gaulin press at 9000 p.s.i.
All procedures were performed between 0 and 4". ?\lembranes were sedimented by a 6-hour centrifugation at 45,000 X g. The supernatant was stored frozen at -20" over a period of months without detectable loss of activity. Streptomycin precipitation of the 45,000 X g supernatant was performed at 0". The frozen material was thawed and diluted I:4 with dist.illed water, yielding a solution (Fraction 1, Table III) of 5 to 10 mg per ml in protein.
A concentrated streptomycin sulfate solution (10%) was added over a period of several minutes with stirring in a volume equal to 15% of that of Fraction 1. After 2 hours at 0", the suspension was centrifuged at 30,000 X g for 20 min.
The streptomycin supernatant, which was free of enzymatic activity, was discarded, and the pellet was redissolved with the aid of a blender in 0.2 M Tris-HCl, pH 7.4, in the same volume as Fraction 1. The redissolved streptomycin pellet (Fraction 2, Table III) was clarified by a second centrifugation at 30,000 x g for 20 min and stored in an ice bath.
To a portion of this fraction, MgC12 (1.0 M) was added to give a final magnesium concentration of 20 mM. Ten per cent sodium deoxycholate was then added to a final level of 0.57,. This mixture was held for la hours at O", and was then centrifuged for 20 min at 30,000 x g. The supernatant (Fraction 3, Table  III)  The protein to RNA ratio of Fraction 4 was about 4f times higher than that of Fraction 1, and about a a-fold increase in specific activity of the enzyme was obtained. However, when Fraction 4 was centrifuged at 100,000 X g, the enzyme was recovered in the precipitate.
When ribosomes were treated with sodium deoxycholate in the absence of magnesium ion, little dissociat.ion of the synthetase from the ribosomes was observed (data not shown), nor does magnesium alone cause dissociation. The formation of a precipitate of magnesium deoxycholate is apparently essential for separation of the enzyme from at, least the bulk of the ribosomal material.
These observations explain why ribosomes prepared by the method of Ron et al. (8) are virtually devoid of phosphatidylserine synthetase.
Formation and Decarbnxylation of Phosphatidylserine in Living Cells-When growing cells of E. coli were labeled with [114C]serine, the lipids of the cell did not retain the carboxyl label unless the decarboxylase was inhibited by hydroxylamine (Fig. 4). Treatment of the cells with hydroxylamine under these conditions did not significantly reduce their viability when subsequently diluted, plated, and counted.
When [3JH]serine was used to label the cells, radioactive lipid accumulated in the hydroxylamine-treated cells at about the same rate as the control, indicating that hydroxylamine acts at the decarboxylase step, and not earlier in the sequence of steps leading to the incorporation of serine into lipid. Thin layer chromatography as described under "Materials and Methods" revealed that the labeled lipid which accumulated in the hydroxylamine-inhibited cells was almost entirely phosphatidylserine.
No radioactive phosphatidglserine could be detected in the lipid extract of the uninhibited cells; the small amount of ra.dioactivity shown in Fig. 4  to some of which NH~OII~IlCl was added in a final concentratibn of 10 mM. Shaking was continued for 10 min, after which DL-[~-I~C] serine was added in a negligible volume at a final concentration of 0.02 mM. The specific activity of the serine was 10' cpm per pmole.
In some samples, the radioactivity was "chased" by the addition of loo-fold excess of unlabeled serine 10 min after t,he labeled serine.
Samples (1.0 ml) were withdrawn at various times before and after chase and mixed with 3 ml of chloroform-methanol (2:1, v/v) containing 0.01 N HCl. The radioactive lipids were extracted and measured as described in the assay of phosphatidylserine synthetase.
to a trace contaminant of the labeled serine, which was, however, 99yc pure by chromatographic analysis. It ma.y be noted that the inhibition of the decarboxylase by hydroxylnmine under these conditions is not complete, since the accumulation of carbosyl-labeled lipid levels off, and the radioactivity is "chased" by the addition of unlabeled serine to the medium. TheFe results show a very rapid decarbosylation of phosphatidylyerine in living cells of E. coli since after as little as 1 min significant amounts of phosphatidylserine more than 1000 counts above the background level had been formed and accumulated in the inhibited cells, but had been completely decarboxylated in t,he uninhibited cells.

Intracellular
Localization of Newly Synthesized Phosphatidylserine-Since phosphatidylserine synthetase and the decarboxylase appear to be localized on different intracellular structures, the experiment of Fig. 4 suggests that a rapid intracellular translocalization could account for the efficient decarboxylation of the newly synthesized lipid.
In an attempt to determine the location of the phosphatidylserine which accumulates in hydroxylamine-inhibited cells, such cells were disrupted and subjected to sucrose gradient centrifugation in the experiment, of Fig. 5. Kearly all of the newly synthesized phosphatidylserine was recovered in the mernbrane fraction, with very little in the region expected t,o contain ribosomes.
The experiment, however, must be interpreted with caution, since redistribution of lipid among membrane fragments may take place during sonic disruption (15). A similar redistribution of phosphatidylserine between ribosomes and membrane fragments may also take place. Even if this is the case, it can be concluded that the equilibrium greatly favors localization of the lipid product on the membrane, under conditions in which the synthetase is largely retained on the ribosomes. A sample (0.4 ml) with a protein concentration of about 0.2 mg per ml was layered on a sucrose gradient, as described in Fig. 1 which also contained 10 mM hydroxylamine throughout.
Fractions from the gradient were extracted with chloroform-methanol, and the lipidsoluble radioactivity was determined.

DISCUSSIOS
In this as in every other study of the intracellular distribution of enzymes, the possibility must. be considered that the pattern observed in cell-free extracts may not correspond wit,h that in the intact, living cell. Redistribution of enzymes may take place during disruption of the cell and handling of the extract?. For example, there is evidence that a ribonuclease often isolated in association with ribosomes is in fact a periplasmir protein which becomes adsorbed to ribosomes during disruption of the cells (16).
RN;\, however, is a substrate for this enzyme, and it can only be activated by procedures which tend to disrupt ribosome structure (17). Neither of these considerations apply to phosphatidylserine synthetase. Furthermore, the apparent ribosomal localization of the enzyme is observed when the much gentler penicillin rnethod is used to prepare cell-free estract,s rather than sonic disruption.
Nevert'heless, the conclusion that the synthetase is a ribosomal enzyme is still subject to the reservations mentioned.
In rat liver, phosphatidylserine is synthesized from phosphatidylethanolamine + L-serine in a reaction catalyzed by an "exchange" enzyme (18). This enzyme is localized in the endoplasmic ret,iculum, whereas phosphatidylserine decarbosylase is ent.irely mitochondrial (19). Nevertheless, L-serine is rapidly converted to lipid and decarboxylated by t,he liver. The important studies of Wirtz and Zilrersmit (20) alld of McMurray and Dawson (21) have revealed the presence in liver of soluble proteins which rapidly carry out the reversible trunslocation of phospllolipids such as lecithin from endoplasmic reticulum to mitochondrial rnernbranes. Such an intracellular transport system would account for the rapid, mitochondrial decarbosylation of phosphatidylserine synthesized on the endoplasmic reticulum. The experirnent of Fig. 4 suggests the possible presence of such translocation proteins in E. coli. We cannot exclude the alternative, however, that the ribosomally bound synthetaae exerts its action directly at the surface of the membrane in zivo. It is also possible that the small fraction of the synthetase not associated with the ribosomes may account for the tot'al activity of the enzyme in living cells, although this seems quite unlikely. The possible significance of the ribosomal localization of phosphatidylserine synthetase deserves some consideration. Phosphatidylethanolamine is the principal membrane lipid of E. coli, and phosphatidylserine synthetase catalyzes the first step in the branch of the pathway leading to phosphatidylethanolamine. It is obviously important for the cell to coordinate the synthesis of protein and of membrane lipids, but the mechanisms by which this is achieved are unknown.
The localization of a key enzyme for the biosynthesis of lipids on ribosomes may be somehow linked to the joint regulation of protein and lipid synthesis.