Choline-containing Teichoic Acid As a Structural Component of Pneumococcal Cell Wall and Its Role in Sensitivity to Lysis by an Autolytic Enzyme*

SUMMARY The cell wall of Diplococcus pneumoniae was isolated and purified and its macromolecular components were characterized in two ways: (a) by selective extractions and (&) by solubilization of the wall with the pneumococcal autolytic enzyme followed by fractionation of the enzyme products. Two major macromolecular components could be identified. One, the peptidoglycan, is composed of glucosamine, muramic acid (plus muramic acid phosphate), lysine, alanine, and glutamic acid in the molar ratios of 1.0: 1.2: 1.5:3.8:2.0. The presence of additional amino acids, namely, aspartic acid, serine, glycine, and threonine in the molar ratios (with respect to glucosamine) of 0.9:1.0:1.0:0.5 was also detected. The other major component, teichoic acid, is rich in galactosamine, phosphate, and choline which occur in the molar ratios 1.0: 1.65: 0.9. These three constituents together make up 23 % of the cell wall mass. treatment cell-free pneumococcal autolytic


TOMASZ
From The Rockefeller University, New York, New York 1005'1 SUMMARY The cell wall of Diplococcus pneumoniae was isolated and purified and its macromolecular components were characterized in two ways: (a) by selective extractions and (&) by solubilization of the wall with the pneumococcal autolytic enzyme followed by fractionation of the enzyme products. Two major macromolecular components could be identified. One, the peptidoglycan, is composed of glucosamine, muramic acid (plus muramic acid phosphate), lysine, alanine, and glutamic acid in the molar ratios of 1.0: 1.2: 1.5:3.8:2.0.
The presence of additional amino acids, namely, aspartic acid, serine, glycine, and threonine in the molar ratios (with respect to glucosamine) of 0.9:1.0:1.0:0.5 was also detected. The other major component, teichoic acid, is rich in galactosamine, phosphate, and choline which occur in the molar ratios 1.0: 1.65: 0.9. These three constituents together make up 23 % of the cell wall mass.
After treatment of cell walls with cell-free pneumococcal autolytic enzyme, most of the cell wall material can be recovered in the form of two soluble macromolecular fractions which are separable by gel filtration.
Fraction I is of high molecular weight and contains teichoic acid polymers plus what appears to be the polysaccharide backbone of the peptidoglycan.
Fraction II contains material of lower molecular weight and is rich in the amino acids lysine, glutamic acid, and alanine; these amino acids appear to represent cross-linked dimers and trimers of the peptide portion of the peptidoglycan.
The teichoic acid and the polysaccharide components of Fraction I can be separated after sequential degradation with nitrous acid and periodate. It has already been proposed in the literature that the antigenic C-polysaccharide of pneumococcus is a teichoic acid.
Our studies further indicate that this teichoic acid contains choline and is a major structural component of pneumococcal cell wall. The major autolytic activity in pneumococcal extracts appears to be an amidase which splits the bond between muramic acid and alanine in the peptidoglycan portion of the cell wall.
Experiments are described which indicate that * This work was supported in part by a grant from the United States Atomic Energy Commission.
A preliminary report of some of this work was published elsewhere (1).
# National Science Foundation Graduate Fellow. the lysis of pneumococci by deoxycholate occurs through the participation of this enzyme. The choline component of the teichoic acid plays a key role in determining sensitivity to the autolytic enzyme, since walls prepared from pneumococci in which cell wall choline was replaced by ethanolamine were found to be totally resistant to the action of the autolytic enzyme.
A novel form of choline in nature was described recently; this substance was identified as a component of some macromolecular structure localized in the cell wall of the bacterium, Dipk~coccus pneumoniae (2).
At least 40 to 50% of the radioactive choline incorporated into the wall from the growth medium could be recovered as a component of the purified C-polysaccharide (2), one of the major antigenic components of pneumococcal cell walls, if the preparative methods of Goebel et al. (3) or Liu and Gotschlich (4) were followed.
A varying fraction of the incorporated choline was also extractable with cold trichloracetic acid (2), a solvent often used for preparing teichoic acids (5) and which has been used to extract material with C-antigen activity from pneumococci (6). A quantitative release of choline in the form of macromolecules could also be achieved by treating live pneumococci with cell-free pneumococcal autolytic enzyme (7).
The physiological importance of the choline-containing macromolecules is indicated by the fact that pneumococci have a nutritional requirement for choline (8). In addition, when the analogue ethanolamine replaces choline in the cell walls of bacteria, the bacteria develop a number of physiological abnormalities; daughter cells do not separate at the end of cell division, the cells lose their ability to undergo genetic transformation, they do not autolyse in the stationary phase of growth, and they become resistant to detergent-induced lysis (9). These observations offer an opportunity to examine the relationship between cell wall primary structure and several properties involving the bacterial surface. Of particular interest is the relationship between wall structure and cellular lysis. A possible role of an autolytic enzyme in detergent-induced lysis has been suggested (10). With these thoughts in mind, we have begun to 287 by guest on March 24, 2020 http://www.jbc.org/ Downloaded from Pneumococci (strain R 3611) were grown in a chemically defined medium (11) buffered at pH 7.6 or 8.0. For the preparation of cell walls containing ethanolamine, choline was replaced in the growth medium with 20 Mg per ml of ethanolamine.
All chemicals used were commercially available analytical grade products. Purchased from New England Nuclear Corporation were 3H-and 14C-labeled radioactive compounds; 32P was obtained from Union Carbide Corporation.
The isotopic tracers were added to the growth medium at 0.1 to 0.5 PCi per ml. The following compounds were used: choline (aH-or 14C-methyl or 1% uniformly labeled), ethanolamine and lysine (both 14C uniformly labeled), and 32P04. Radioactive samples were either dried on glass fiber filters (Whatman, GFA, 2.4 cm) and placed in toluene scintillation fluid or liquid samples were added to ethanol-toluene scintillation fluid or Bray's solution and counted in a scintillation spectrometer (Nuclear Chicago Unilux or Mark I). Double label counting of 3H and 14C was performed by the discriminator-ratio method (12).
Sephadex gels were obtained from Pharmacia. All columns were prepared as recommended, and void volumes were determined using solutions of blue dextran.
The conditions for the individual cases are described in the legends to figures.
Preparation of Cell Walls-Cell walls were prepared from exponentially growing cells according to a modification of a published procedure (2). The cells were washed two to four times (centrifugation, 12,000 X g, 10 min) with 0.15 M NaCl (in about 0.01 of original culture volume) and resuspended in the same volume of water (or occasionally in 0.15 M NaCl).
After heating at 6570" for 15 min to inactivate the autolytic enzyme, the cells were washed once, resuspended in water, and disrupted by shaking in a Mickle cell disintegrator or in a Braun tissue homogenizer with an equal volume of washed glass beads (Ballotini No. 13) for 15 min, a treatment which left no recognizable bacterial forms. The broken cells were separated from the glass beads and the suspension was centrifuged at low speed (3,000 X g, 5 min) to remove any unbroken cells. The walls were then washed twice with acetone, twice with 0.15 M NaCl containing 0.01 M K2HP04 (saline-phosphate buffer) and resuspended in the same. DNase (10 pg per ml) and RNase (50 pg per ml) (Worthington Biochemicals) and a few drops of chloroform were added and the suspension was incubated at 37" for 4 to 12 hours with constant mixing with a magnetic stirrer.
Trypsin, 50 pg per ml (Worthington Biochemicals), and 0.1 mM CaC& were added and incubation was continued for 12 hours. The suspension was then incubated with a second portion of trypsin for an additional 12 hours. The cell walls were washed six times in saline-phosphate buffer and six times in water. This number of washes was sufficient to reduce the counts released from isotope-labeled cell walls virtually to zero. Preparation of Autolytic Enzyme-Exponentially growing cells were harvested by centrifugation (as above), washed and resuspended in 0.01 volume of saline-phosphate buffer containing 0.002 M 2-mercaptoethanol.
To a suspension of 5 X lOlo bacteria per 5 ml, deoxycholate (0.03 to 0.1%) and pancreatic DNase (0.5 to 1 pg per ml) were added. The suspension was incubated at 37" until maximum clearing occurred, usually in 10 to 15 min. The clarified suspension was centrifuged (12,000 X g, 10 min) to remove large particulate matter. The supernatant was passed through two glass fiber filters on top of a Millipore filter (type HA, 2.5 cm) held in a Millipore filter apparatus. The filtrate was filtered through a Sephadex G-75 column (3 x 40 cm, eluent salinephosphate buffer containing 0.002 M 2-mercaptoethanol) in the cold to remove deoxycholate.
"Autolytic" enzymatic activity eluted in the void volume of the column.
The active fractions could be stored at 4" for several months with no detectable loss of activity.
Autolytic Enzyme Assay-An 0.1 ml aliquot of the enzyme preparation (containing about log cell equivalents of bacterial lysate) was added to 0.9 ml of a suspension of (saline-phosphate buffer containing 0.002 M 2-mercaptoethanol) cell walls (0.1 to 50 mg, dry weight, prepared from bacteria grown on choline) labeled either with radioactive choline or with radioactive lysine or both. After incubation at 37", samples were removed and centrifuged (International Micro-Capillary centrifuge, model MB, 14,000 x g, 10 min) and the amount of released radioactive label was determined.
Occasionally, when enzymatic hydrolysis was carried out on a large scale with more concentrated suspensions of cell walls, a precipitate formed during the course of the reaction. The choline-and lysine-labeled components of the cell wall were always quantitatively released to the soluble fraction. Nitrous Acid and Perioda?e Treatment-To a l-ml aqueous solution of the choline-labeled fraction of the enzyme products from about 35 mg of cell walls, were added 1 ml of 5 y0 NaNOz and 1 ml of 5.8 N acetic acid (13). The solution was incubated at 25" for 20 min and then adjusted to pH 5 to 6 by the addition of Na2C02 solution.
The extent of fragmentation was ascertained by filtering the treated material through a column (1 X 50 cm) of Sephadex G-100 with water as the eluent.
After assaying aliquots of each fraction, all fractions were pooled and concentrated by lyophilization.
This material, as well as the untreated choline-containing enzyme products, was treated with periodate as follows.
To about 1 ml of aqueous solution of the sample containing 10 to 20 mg of material was added an equal volume of 0.1 M sodium metaperiodate, and the solution was incubated at 37" for 1 hour in the dark at pH 7. Unreacted periodate was destroyed by the addition of glycerol or ethyleneglycol and incubation for an additional hour.
Periodate reduction products were removed by filtration through a column (1 X 50 cm) of Sephadex G-10 with water as the eluent; the fragmented enzyme products were excluded from the gel, while the reduction products of periodate were retained.
The treated sample was concentrated by lyophilization. Dinitrophenylation was carried out by a standard procedure 04).
Analytical Procedures-Choline was estimated by bioassay using Neurospora craSSa as the test organism (15). Samples were hydrolyzed for 12 to 16 hours in 6 N HCl, flash evaporated, and the residues were taken up in water for assay. Phosphate was determined by the method of Chen, Toribara, and Warner (16). Amino sugars and amino acids were determined with a Beckman model 120B automatic amino acid analyzer. Samples were hydrolyzed in 6 N HCl in tubes which had been sealed under vacuum.
Cell walls were hydrolyzed at llO", while other samples were hydrolyzed at 100". After hydrolysis, samples were flash evaporated and the residue was taken up in 0.2 N citrate buffer (pH 2.2) for analysis.
Decomposition curves for the amino sugars were determined separately for each material analyzed, since quite marked variations in decomposition rates in different materials were observed. Glucosamine, muramic acid,   (17) and for which we have used the color constant experimentally determined for muramic acid.) Since the amino sugars and amino acids in similar material have been identified (4), we have not isolated the individual components from the amino acid analyzer column to check their identity.

Distribution
of Choline-containing Molecules during Cell Wall Preparation-During the preparation of cell walls from cells labeled with radioactive choline, it became apparent that not all the choline which had been incorporated by cells remained associated with sedimentable wall structures.
In order to determine if the choline released during the different steps of cell wall preparation represented specific subfractions of the choline-containing molecules, we measured the amount released by each treatment and examined their elution properties on Sephadex G-100 (Table I and Fig. 1). VVe have included in the table similar measurements of the release of radioactive lysine from choline-grown cells and the release of radioactive ethanolamine and lysine from ethanolamine-grown cells (9). Note that about 70% of the ethanolamine but only 40% of the choline originally associated with the cell suspension remained in the final wall preparation.
It is possible that the action of endogenous autolytic enzyme or enzymes during harvesting of the bacteria contributes to the loss of choline from the cell wall.
The gel filtration properties of choline-labeled material released in the various steps of the wall preparation ( Fig. 1) suggest that they may represent distinct classes of molecules.
These fractions have not been further characterized, but it is likely that they contain biosynthetic intermediates,' as well as breakdown products, of wall polymers.
One must therefore bear in mind that isolated cell walls probably do not contain a complete complement of all the choline-containing molecules of the cell. Extraction of Choline-containing Molecules from Cell Walls-Purified cell walls were treated with reagents previously used either to characterize or to extract various types of bacterial cell wall polymers or to do both (Table II).
The choline-labeled molecules are solubilized to a much greater extent than the lysinelabeled ones by all the solvents which are known to be selective extractants of either teichoic acids or polysaccharides or both. Much of the lysine label was solubilized by 10% trichloracetic acid, and neither label was extracted by phenol.
The gel filtration properties of the extracted choline label (Fig.  2) show that, with the exception of hydrochloric acid, the treatments released choline still attached to large molecules.
The formamide-treated material is completely excluded by Sephadex G-100 (Fig. 2B). Fragmentation of the choline-containing polymers by all the other treatments is apparent.
Composition of Pneumococcal Cell Walls-The composition of cell walls prepared by the methods described above is given in Column 1 of Table III.
Molar ratios of the components present are expressed with respect to galactosamine.
The constituents listed account for 70.3 To, dry weight, of the preparation. Fig. 3 is an outline of the scheme used to solubilize and fractionate cell wall polymers.
Each step is described in detail below. Hydrolysis of Cell Walls with Autolytic En.zyme-Both choline and lysine labels are completely solubilized from purified cell walls 1 Substantial quantities of phosphorylcholine have been identified in the supernatant fraction of mechanically disrupted cells (unpublished data).
Cel l Wal l s (  cell walls were hydrolyzed with the autolytic enzyme (total of 5 x log "cell equivalents"; see "Experimental Procedure") in a volume of about 150 ml. At the indicated times, aliquots of the reaction mixture were removed, heated at 65' for 10 min to inactivate the enzyme, centrifuged, and nonsedimentable aH and 14C counts were determined.
A precipitate which formed during the course of hydrolysis under these conditions was removed by the centrifugation.
Choline-and lysine-labeled components were exclusively in the soluble fraction.
The choline-and lysine-labled products can be readily separated on Sephadex G-25 (Fig. 4B). Only two peaks are obtained. Peak I contains all the choline label and is excluded from the matrix of the gel. Peak II contains only lysine label and is retained by the gel. A fraction of the lysine label varying in quantity from experiment to experiment elutes in Peak I, probably reflecting different extents of hydrolysis.
The main part of Fig. 4B is a plot of the results from Experiment 3. For convenience, Peaks I and II will be referred to as the "choline-labeled" and "lysine-labeled" enzyme products, respectively. Characterization of Lysine-labeled Enzyme Products--After concentration of the lysine-containing enzyme products by lyophilization, the material was passed through gel filters in order to obtain an estimate of its molecular weight.
Sephadex G-10 and G-15 and Bio-Gel P-4 columns were used, with 1 M sodium chloride solution as eluent to minimize adsorption to the gels. The results of these experiments (Fig. 5) indicate an approximate molecular weight between 700 and 1500.
The amino sugar and amino acid composition of this material was determined with an automatic amino acid analyzer after acid  were detected. The amino acids lysine, alanine, and glutamic acid are the major components of this fraction, representing 85% of the amino acids detected. The molar ratios obtained were lysine-alanine-glutamic acid, 1 .O : 2.1 :l.l. This is similar to the ratio of the same amino acids in whole cell walls (see Table III, Column l), 1.0:2.5 : 1.3. Of the other seven amino acids detected in this fraction, the most abundant ones were aspartic acid, serine, and glycine. They were present in the ratios 0.1:0.3 :0.2 per lysine residue.
Up to 86% of the radioactive lysitie associated with the purified cell walls was recovered in this fraction (see Fig. 4). Choline-containing Enzyme Products- Fig.  6 shows the elution profile from Sephadex G-100 of the choline-containing polymers released from purified cell walls by autolytic enzyme treatment. In three of four such preparations, most of the material was excluded from the Sephadex G-100 gel; in one preparation, however, part of the material appeared fragmented to smaller sizes. Heterogeneity of size and of ionic character is indicated by the results of chromatography on Sephadex G-200 dextran gels and on DEAE-Sephadex (Fig. 7, A and B). This fraction of the enzyme products represented 44.6% of the mass of the starting cell walls. Choline, phosphate, amino sugar, and amino acid analyses were performed on this mat.erial.
About 74yo of the dry weight could be accounted for in terms of the components listed in Table III, Column 2.
Pertinent data for the average composition of C-polysaccharide (4) have been included for comparison (Table III, Column 5). Although Liu and Gotschlich (4) did not analyze for choline in their preparations, we have subsequently determined the amount of this compound in a C-polysaccharide preparation provided by Dr. Emil C. Gotschlich and obtained a value of 0.74 f 0.09 pmole per mg. We have included this value in Table III. By comparing the composition of the enzyme products (Table  III, Column 2) with that of the starting cell walls (Table III, Column l), the enzyme products are seen to be enriched in choline, phosphate, galactosamine, glucosamine, muramic acid and mura-mic acid phosphate, and aspartic acid. Between 60.9 and 72.2% of the amounts of these components in the starting cell walls was recovered in this fraction with very little change in the molar ratios. Relative to the starting cell wall material, the concentration of all amino acids except aspartic acid is decreased in the enzyme products; only 18 to 44y0 of the various amino acids remained in this fraction and their concentration could probably be further decreased by more extensive autolytic enzyme treatment. A decrease in lysine, alanine, and glutamic acid is expected, since about 60 to 80% of these amino acids are recovered in the lysinelabeled fraction of the enzyme products.
Of the amino acids remaining in the choline fraction, glutamic acid and aspartic acid show an increase and alanine a decrease relative to lysine. Thus, the ratio of lysine-alanine-glutamic acid-aspartic acid is 1.0: 1.6: 1.7:1.3 compared to 1.0:2.5:1.3:0.6 in the starting cell walls. The molar ratios of the other amino acids are not changed significant,ly. If one aSsumes that all the lysine and alanine could be obtained, under ideal conditions, in the lysine-labeled fraction, there would still remain an excess of 0.9 glutamic acid and 1.2 aspartic acid residues per lysine residue.
We cannot account for the loss of the amino acids from valine downward in the table. Since only 10 to 20% of these particular amino acids are recovered in the lysine-rich fraction, it seems that they are removed either in the precipitate which forms during enzymatic hydrolysis of the walls or during the Sephadex G-25 or G-100 fractionation steps. Another conclusion from these analyses is that the average amino acid and amino sugar composition of the enzyme products is quite similar to that of Liu and Gotschlich's C-polysaccharide (   IO  20  30  40  50  30  60  90  120  0  30  60  90  30  60  0  30  60  90  0 3o" Fraction number the cell walls derived from them (Fig. 2C), or the choline-labeled fraction of the autolytic enzyme products (Fig. 8A). Treatment with nitrous acid also causes fragmentation of the choline-containing polymers and periodate treatment applied after nitrous acid causes still further fragmentation (Fig. 8, B  FIG. 8. Sephadex G-25 elution profiles of choline-labeled fragments prepared by treating the choline-labeled fraction of the autolytic enzyme products with sodium periodate (A), nitrous acid (B), and nitrous acid followed by sodium periodate (C). A, periodate-treated 14C-choline labeled enzyme products; B, nitrous acid-treated enzyme products labeled with 3H-choline and 1%. lysine (the starting material is shown in Fig. 6) ; C, the material Combination of these treatments was used for the further fractionation of the choline-labeled enzyme products. After treatment with 0.24 M nitrous acid and then with 0.05 M sodium metaperiodate (see Fig. 3), filtration of the reaction mixtures through Sephadex G-25 yielded two subfractions (Fig. 8C) : (a) a cholinelabeled fraction which was retained by the column and (5) a fraction which was excluded.
The position of the latter fraction in the elution profile was marked by a small peak of i4C-lysine in the exclusion volume of the column.
A total of 92% of the dry mass of the starting material was recovered in these two fractions: 66.5% in the choline-rich fraction and 25.7% in the lysine-rich part.
The composition of the two fractions is shown in Table III. Fraction I (Column 3) contains 58.3% of the choline, 47.3% of the phosphate, 48.2% of the galactosamine, and 35.1 y0 of the muramic acid (and muramic acid phosphate) of the cell wall and represents about 75% of the mass of choline-labeled enzyme products. The molar ratios of choline, phosphate, and galactosamine are changed very little from that characteristic of purified cell walls, whereas the relative amounts of glucosamine, muramic acid, and of all the amino acids are drastically reduced.
The proportions of peptidoglycan components remaining in this fraction are quite different from those characteristic of peptidoglycan (21). For example, the ratio of glucosamine to muramic acid plus muramic acid phosphate is 1.0:3.5. The relatively small amounts of amino acids left in this fraction have the proportions of 1.0 lysine, 2.5 alanine, 4.3 glutamic acid, 2.8 aspartic acid, 1.2 threonine, 1.2 serine, and 2.2 glycine.
The composition of Fraction II (Column 4, Table III) suggests that it consists primarily of molecules derived from the polysaccharide backbone of the peptidoglycan with some of the original complement of amino acids attached.
It is apparent that the proportions of these substances also deviate from those normally found in peptidoglycan.
For example, the ratio of glucosamine to muramic acid plus muramic acid phosphate is 1 .O : 1.5, and that of lysine to alanine to glutamic acid is 1.0:3.0:3.4.
In spite of the fact that 92.2% of the weight of the starting material was recovered in the two fractions prepared with nitrous acid and periodate, there was substantial loss in the recovery of several specific components.
For instance, only 38% of the lysine was recovered and only 70% of the input glucosamine could be accounted for. It seems most likely that the apparently distorted proportions of peptidoglycan components observed in shown in B after subsequent periodate treatment. The same column (1 X 50 cm) was used in B and C; another column (1 X about 40 cm) was used in A. Fraction volumes were about 1 ml. A, l represents radioactivity (14C-choline) ; B and C, 0 indicates 3H (choline) and 0 represents 1% (lysine).
The optical density of blue dextran is indicated by f. these fractions result from destruction of some residues by nitrous acid or sodium periodate, and do not represent an unusual assembly of cell wall components.
The fact that only 56% of the weight of the choline-containing material (Table III, Column 3) can be accounted for by the components listed indicates this also. This unaccounted for weight could represent degradation products of ribitol, a diaminotrideoxyhexose, and glucose (6) (see below).
Deoxycholate-induced Lysis and Autolysis- Fig.  7 shows the chromatographic and gel filtration properties of (a) choline-containing material prepared from purified cell walls with autolytic enzyme (Fig. 7, A and B) and (5) the choline-containing material prepared by deoxycholate lysis of whole cells (Fig. 7, C and D). There are several points to be made from this figure. We have already mentioned the heterogeneity of the choline-labeled enzyme products obtained from purified cell walls. The elution profiles in Fig. 7, A and B, were obtained with enzyme products prepared from cell walls doubly labeled with 3H-choline and 32P04. The almost perfect overlap of the 3H and 32P in the elution profiles indicates that the isotopes label the same molecules.
The elution profiles in Fig. 7, C and D, were obtained with a deoxycholate lysate of 14Ccholine labeled cells. Although C and D are not identical with A and B, respectively, they are quite similar.
Furthermore, the elution profiles of C-polysaccharide isolated from a deoxycholate-induced lysate on Sephadex G-200 and DEAE-Sephadex under the same conditions (22) are quite similar to Fig.  7, A and B, respectively.
These findings, together with the close similarity in chemical composition of our choline-containing enzyme products and that of C-polysaccharide (prepared and analyzed by Liu and Gotschlich (4)), indicate a basic similarity between deoxycholate-induced lysis and the hydrolysis of purified cell walls by the isolated autolytic enzyme. This similarity suggests that deoxycholate-induced lysis can be explained by the action of an endogenous autolytic enzyme, as originally proposed by Dubos (10).
Treatment of Ethanolamine-containing Cell Walls with Autolytic Enzyme-In view of the similarity between deoxycholate-induced lysis and the lysis of purified cell walls by cell-free enzyme preparations, it was of particular interest to test the sensitivity of purified ethanolamine-containing cell walls to cell-free enzyme. It will be recalled that bacteria containing ethanolamine in their cell walls become resistant to deoxycholate-induced lysis (9). Cell walls labeled with 14C-ethanolamine and with 14C-lysine were pre- pared from bacteria grown in ethanolamine medium, and enzymatic hydrolysis of these walls was attempted using the standard assay procedures. Table IV shows that neither the ethanolamine nor the lysine lable is solubilized from the ethanolaminecontaining walls under conditions in which 3H-choline and 1%lysine are essentially completely released from cell walls containing choline.

Composition of Pneumococal Cell WallsTo
our knowledge, the cell walls of D. pneumoniae have not been characterized before. The walls contain large amounts of choline, phosphate, and galactosamine.
The presence of somewhat more than 1 lysine and glutamic acid and more than 2 alanine residues per muramic acid indicates that pneumococcal peptidoglycan contains excess peptide units and is therefore probably an example of the type III peptidoglycan of Ghuysen (23). The amino acids aspartic acid, threonine, serine, and glycine are also present in our wall preparations in the proportions of 0.9 aspartic acid, 0.5 threonine, 1.0 serine, and 1.0 glycine per glucosamine residue. In addition, a number of amino acids not typical of peptidoglycan are also present in significant quantity.
As to the origin of the "non-peptidoglycan" amino acids in our cell wall preparations, although cytoplasmic contamination cannot be rigorously excluded, it seems unlikely from the following considerations.
In preparations of cell walls from 14C-lysine-labeled cells, wall-bound lysine label can be reduced to an apparently minimum level, after which further trypsin treatment and washing cause no more loss of counts.
Moreover, at least 86 y0 of the lysine label remaining with the cell walls can be obtained in the form of small molecules with a composition characteristic of the peptide portion of peptidoglycan.
Thus, if the non-peptidoglycan amino acids are present in protein, that protein is deficient in lysine, alanine, and glutamic acid and therefore would not be indicative of cytoplasmic contamination.
Another consideration is that the non-peptidoglycan amino acids which could be quantitated in the fractions prepared from cell walls (Table III), valine, leucine, and isoleucine, show little change in their molar ratios with respect to each other, although as a group their ratios with respect to galactosamine, for example, differ from fraction to fraction. It is interesting to speculate in this connection that these amino acids may be components of a wall protein such as an autolytic enzyme.
Fragmentation of Cell Walls-Dissolution of purified cell walls by treatment with autolytic enzyme and the separation of the solubilized products by gel filtration permit the recovery of about 72% of the original dry mass of cell wall material in the form of two main fractions: Fraction I, a large molecular weight, cholinerich fraction and Fraction II, a smaller molecular weight fraction rich in lysine, alanine, and glutamic acid. Extraction studies and chemical analyses indicate that Fraction I contains mainly components of a teichoic acid-like polymer plus the polysaccharide backbone of the peptidoglycan, whereas the size and composition of Fraction II show that it represents the peptide portion (cross-linked dimers or trimers of a tetrapeptide containing 1 lysine, 1 glutamic acid, and 2 alanine residues) of the peptidoglycan. The amount of peptide remaining in Fraction I varies from experiment to experiment but at least 86% of the peptide can be removed from this fraction.
Interestingly, aspartic acid, serine, glycine, and threonine remain with Fraction I.
Fraction I can be further resolved by degradation first with nitrous acid and subsequently with periodate.
About 92% of the mass of this fraction can be recovered after treatment with these reagents in the form of two subfractions: one which is retained by Sephadex G-25 and contains all the choline and another which is excluded by Sephadex G-25 and contains glucosamine and muramic acid. Thus this procedure seems to achieve the separation of a teichoic acid-like component (in degraded form) from the polysaccharide backbone of the peptidoglycan.
Nature of Macromolecular Choline Carrier-One of the main purposes of this study was to identify the macromolecular structure or structures to which choline is attached.
It was already known from earlier work that choline is attached to some structure localized in the anatomical region of cell walls (2). The studies described here further specify this structure as a polysaccharide or teichoic acid which is rich in galactosamine and phosphate as well as choline, and which is a structural part of the pneumococal cell wall.
This classification of the macromolecular carriers of choline is based on the following observations.
(a) Macromolecular substances rich in choline can be selectively extracted from purified cell walls (without solubilization of the peptidoglycan) by reagents which are known to extract teichoic acids and polysaccharides; (b) treatment of purified cell walls with autolytic enzyme prepared from pneumococci can solubilize the choline-containing macromolecules substantially "freed" of the peptide component of the cell walls; (c) recovery of the choline label after various degradative (enzymatic and chemical) treatments of cell walls invariably yielded molecules which also contained galactosamine and phosphate in addition to choline. The molar ratio of these components remained essentially unchanged in spite of the fact that these treatments caused drastic changes in molecular size; (d) the presence of choline and phosphate in the same molecules is also indicated by the cofractionation of %choline and 32P04 label during ion exchange and Sephadex G-200 chromatography of the autolytic enzyme products.
C-Polysaccharide and Choline-containing Teichoic Acid-There are two series of recent publications which are relevant to our findings. Liu and Gotschlich (4,22) and Brundish and Baddiley (6, 24) have isolated and analyzed the C-polysaccharide from Teichoic Acid and Lysis Xensitivity Vol. 245,No. 2 pneumococci. The presence of covalently linked choline in C-polysaccharide prepared by the method of either Liu and Gotschlich (4) or Brundish and Baddiley (6) m as reported earlier (2). Recently, the latter authors (6) have confirmed the presence of choline in C-polysaccharide prepared by repeated extraction (for a week) of defatted, dried pneumococci with cold trichloracetic acid followed by purification with Cetavlon and anion exchange chromatography.
Such preparations had high antigenie activity, were essentially free of amino acids, and contained choline, galactosamine, and phosphate in the molar ratio of 1.0:1.0:2.1, similar to the ratios of these same components (0.9: 1.0: 1.6) in our cell wall preparations and in our cholinerich autolytic enzyme products.
The material analyzed by these authors (6) contained, in addition, about 1 residue each of ribitol and glucose and probably about 1 residue of a diaminotrideoxyhexose per choline residue. We detected a faint spot corresponding to the dinitrophenylated derivative of diaminotrideoxyhexose upon electrophoresis of an acid hydrolysate of I-fluoro-2,4-dinitrobenzene-treated cell walls (25). Furthermore, the direct Ehrlich test applied to periodate fragments of the teichoic acid gave strong positive reactions2 (25). On the basis of these observations, it is likely that the diaminotrideoxyhexose is also present in the t-ichoic acid component of our wall preparations. This conclusion is supported by the fragmentation of the teichoic acid by nitrous acid; Brundish and Baddiley (6) concluded that nitrous acid reacted with the diaminotrideoxyhexose in their trichloracetic acid-extracted C-polysaccharide causing fragmentation of the teichoic acid polymer.
We did not analyze our cell wall preparations for ribitol and glucose.
Liu and Gotschlich prepared their C-polysaccharide by a modification of the original procedure of Tillett, Goebel, and Avery (26); pneumococci were lysed with deoxycholate and the lysate was purified by fractional alcohol precipitations, treatment with nucleolytic and proteolytic enzymes, and extensive dialysis.
Such preparations had high antigenic activity and contained covalently bound amino acids, glucosamine, and muramic acid as well as galactosamine and phosphate. Our own analyses on such preparations demonstrated the presence of choline.
The close similarity, in composition and in fractionation properties, between such C-polysaccharide preparations and our choline-rich autolytic enzyme products has already been pointed out. The choline to phosphate ratios are very similar, being 0.59 in our material and about 0.5 in the C-polysaccharide, although with respect to the amount of galactosamine present, the values for choline and phosphate in C-polysaccharide are somewhat lower.
Such variations could be due to differences in the growth medium of the bacteria from which the materials were prepared.
Alternatively, some selective loss of choline and phosphate from the polymeric material may be caused by a drastic early step (heating at 100" at pH 4) of the procedure used to prepare C-polysaccharide (4). Another difference is the smaller amounts of amino acids in the C-polysaccharide.
However, as we have stated before, it is likely that the amount of lysine, alanine, and glutamic acid could be further reduced in our material by more extensive hydrolysis.
We cannot explain the difference in the aspartic acid content of the two materials.
On the basis of the foregoing discussion it seems quite likely that the two kinds of C-polysaccharide preparations, the one 2 A. Tomasz, unpublished data.
free of amino acids and the other containing some components of the peptidoglycan, are both derived from pneumococcal cell walls in which the teichoic acid and peptidoglycan chains form a single, covalently interlinked network. Apparently breaks can be introduced into this network at a number of different sites resulting in the release of the antigenically active galactosaminerich macromolecules (22) in covalent association with varying portions of the other polymers of the cell wall. Thus cold trichloracetic acid seems to release teichoic acid substantially free of peptidoglycan.
It should be added that this procedure also seems to introduce breaks in the backbone oE teichoic acid, and, in our hands at least, trichloracetic acid extraction never solubilized more than about 50% of choline from mid log phase bacteria.
On the other hand, in deoxycholate-induced lysis, teichoic acids seem to be solubilized by autolytic enzyme or enzymes which split bonds between the peptide side chains and the polysaccharide backbone of the peptidoglycan, releasing high molecular weight complexes of teichoic acid and peptidoglycan backbone with some residual amino acids still attached.
It is interesting to point out that the presence of ribitol, glucose, and diaminotrideoxyhexose in th? cell wall in approximately the proportions suggested by Brundish and Baddiley (6) wouid enable us to account for almost all of the dry weight of the wall.
For example, the presence of 1 residue each of ribitol, glucose, and diaminotrideoxyhexose per 2.1 phosphate residues or per 1.0 galactosamine residue (which represent the limits of the disparity between results) would bring the percentage of the weight of the wall accounted for up to 89.4y0 or 94.9%, respectively.
Similarly, the prezence of these component,s in the same proportions in the choline-labeled fraction of the enzyme products brin ;s the respective percentages of weight accounted for up to 100.8 or 114.3%.
This seems also true for the C-polysaccharide preparations of Liu and Gotschlich. According to this interpretation, therefore, one should regard the teichoic acid components in the trichloracetic acid extracts of Brundish and Baddiley (6), in the material prepared from deoxycholate lysates by Gotschlich and Liu (22) and in our cell walls to be essentially identical.
Gotschlich and Liu, however, proposed that the structllre of the teichoic acid in their preparations was poly-N-acetylgalactosamine phosphate, which is not compatible with the above interpretation. The proposal by Gotschlich and Liu was based principally on two observations. (a) A polymer containing N-acetylgalactosamine and phosphate was isolated by cold trichloracetic acid extraction of the "detritus" of a pneumococcal autolysate.
The material had C-antigen activity and 86% of its dry weight could be accounted for as N-acetylgalactosamine and pho ;phate, a composition clearly not permitting even an additional mole of choline per mole of galactosamine.
A similar polymer in which 95% of the ninhydrin positive material was accounted for by galactosamine was extracted with cold trichloracetic acid from purified C-polysaccharide.
(b) Gotschlich and Liu further reported loss of antigenie activity (measured by precipitin assay) and destruction of galactosamine residues by periodate oxidation of these polymers.
Although it is possible that polymers of N-acetylgalactosamine phosphate exist in pneumosocci, we would suggest that, instead of being typical pneumococcal teichoic acids, .uch molecules represent a special, minor class of wall components or that they may arise during the special conditions of prolonged toluene autolysis. It would be important to know what percentage of the total wall material such "86 T0 pure" N-acetylgalactosamine phosphate polymers represent.