Characterization of polynucleotide phosphorylase from Micrococcus luteus and isolation of the 13,000 base poly(A) product of the polymerization reaction.

A new purification procedure for polynucleotide phosphorylase from freeze-dried Micrococcus luteus cells gives approximately 20% yield of nearly homogeneous, primer-independent enzyme which is free of nucleic acid. The physicochemical properties of M. luteus polynucleotide phosphorylase are similar to those previously described for the enzyme from Escherichia coli in terms of Mr, subunit structure, and amino acid composition. The purified enzyme appears to be a trimer composed of three identical subunits (Mr 92,000), but it probably does not exist as such in the cell. Ferguson plot analyses of enzyme in cell extracts indicate that prior to purification the enzyme exists in oligomeric forms characterized by both higher charge and greater Mr. Changes in size and charge of oligomers which occur during purification are probably due to the dissociation of proteins and/or nucleic acids. Dissociation of the oligomers is achieved by dilution and electrophoresis, but reassociation does not occur after concentration. The poly(A) product of the initial polymerization stages migrates as a single band on both nondenaturing and urea-agarose gels. It is 13,000 +/- 2,000 nucleotides long, as measured by electron microscopy, and 8,000 nucleotides long by gel electrophoretic analysis. This poly(A) product remains bound to the enzyme after synthesis, yet can be easily obtained free of protein by proteinase K digestion.

A new purification procedure for polynucleotide phosphorylase from freeze-dried Micrococcus luteus cells gives -20% yield of nearly homogeneous, primerindependent enzyme which is free of nucleic acid. The physicochemical properties of M. luteus polynucleotide phosphorylase are similar to those previously described for the enzyme from Escherichia coli in terms of Mr, subunit structure, and amino acid composition.
The purified enzyme appears to be a trimer composed of three identical subunits (Mr 92,000), but it probably does not exist as such in the cell. Ferguson plot analyses of enzyme in cell extracts indicate that prior to purification the enzyme exists in oligomeric forms characterized by both higher charge and greater Mr. Changes in sue and charge of oligomers which occur during purification are probably due to the dissociation of proteins and/or nucleic acids. Dissociation of the oligomers is achieved by dilution and electrophoresis, but reassociation does not occur after concentration.
The poly(A) product of the initial polymerization stages migrates as a single band on both nondenaturing and urea-agarose gels. It is 13,000 f 2,000 nucleotides long, as measured by electron microscopy, and 8,000 nucleotides long by gel electrophoretic analysis. This poly (A) product remains bound to the enzyme after synthesis, yet can be easily obtained free of protein by proteinase K digestion.
Polynucleotide phosphorylase from Micrococcus luteus has been isolated in two forms: a primer-independent form in which polymerization activity is stimulated only 1.1-to 2-fold by small oligonucleotides (Form I), and a primer-dependent form generated by partial proteolysis in which polymerization activity is stimulated up to 20-fold by oligonucleotide primer (Form T) (1-5). Because of its key role in the synthesis of model nucleic acids (6, 7), there is a need for a preparative procedure which provides enzyme of defined structure and function, free of contaminating activities. However, in procedures published to date, both the yields of enzyme and degree of proteolysis (and thus primer stimulation) have been variable, primarily due to an extreme sensitivity to proteolytic * 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.
f Present address, National Institute of Neurological and Communicative Disorders and Stroke, National Institutes of Health, Bethesda, MD 20205. breakdown. This has kept the structure of M. luteus polynucleotide phosphorylase surrounded by uncertainty as exemplified by multiple electrophoretic polypeptide bands corresponding to a size range from 59,000 to 70,000 for the primerdependent enzyme and 67,000 to 100,000 for the primer-independent form (2,8). With the present procedure, we are able to reproducibly isolate enzyme in milligram amounts (8 mg/ 100 gm of freeze-dried cells) with minimal proteolysis as evidenced by gel electrophoresis. This has allowed us to conduct studies on the chemical and physical characterization of intact M. luteus polynucleotide phosphorylase and to examine the nature of higher level aggregates of the enzyme apparently present in the cell. The structure of the M. luteus enzyme is very similar to that of the enzyme from Escherichia coli. However, the poly(A) synthesized by the M. luteus enzyme is large (-13,000 nucleotides by electron microscopy) compared to the previously reported values of 200 to "greater than" 700 bases for M. luteus poly(A) (4, 9) or 200 bases for the poly(A) synthesized by the E. coli enzyme (10).

Properties of Purified Polynucleotide Phosphorylase
The purification procedure described in the miniprint section gives polynucleotide phosphorylase with a specific activity of 45 units/mg (see footnote to Table I in the miniprint) in a yield of 18%. The enzyme is primer-independent, showing only a 1.1-to 2-fold stimulation when assayed for polymerization in the presence of (AP)~A. Analysis of the protein by SDS gel electrophoresis had shown (11) that 90% of the protein co-migrated with the u subunit of RNA polymerase creased in order to spread the gel pattern across the entire gel length (buffer system 2860.15.VII.) (see "Methods"), and resolution was thereby sharpened, there were two major bands with M, = 220,000 and 200,000, respectively. These are "size isomers" (19), i.e. they differ in size rather than net charge, since they share a common y-intercept (YO) on a Ferguson plot (see below).
This heterogeneity of the purified enzyme is seen in the activity and protein staining patterns in gel electrofocusing (Fig. 1). Both activity (Fig. 1A) and protein patterns (Fig. 1,  B to E ) show a dominant broad zone reminiscent of the "sets" of enzyme apparent in gel electrophoresis. These zones can extend over 2 pH units, and even at the lowest load (Fig. lE), the major band extends over an entire pH unit (6-7). These patterns exhibit vertical streaking indicating insolubility of the enzyme near its isoelectric position. This streaking is independent of protein load between 6 and 60 pg. In addition, the protein patterns, particularly those of overloaded gels, show minor contaminant bands. When isoelectric focusing was done in the presence of 8 M urea (Fig. lF), the enzyme was resolved into a major band a t pH 6.1 and a minor band at pH 6.8. (The ratio of the two components corresponds approximately to that of the two subunits detected by SDS gel electrophoresis (1 1) .) Physical and Chemical Properties of Polynucleotide Phosphorylase-The molecular weight of the purified enzyme was determined by sedimentation equilibrium as shown in Fig. 2. It was found that the best fit was obtained by assuming two noninteracting components of M, values of 237,000 f 24,000 and 830,000 f 200,000, respectively. 92% of component 1 and 8% of component 2 are present. The errors cited are those attributable to fitting and do not take into account possible errors in molecular weight determination due to uncertainties in the value of the partial specific volume.
The molecular weight of the major component suggests that the enzyme in the native conformation is a trimer. This was c o n f i e d by SDS gel electrophoresis on the enzyme crosslinked with dimethylsuberimidate according to Davies and Stark (25). The three components seen upon gel electropho--6.8 Isoelectric focusing of purified polynucleotide phosphorylase (Fraction VII). Gels A to E were subjected to isoelectric focusing under nondenaturing conditions. Gel A (10 pg) was electrofocused for 24 h (see "Methods") and stained for activity with ethidium bromide (61). Gels E to E (gel E , 65 pg; gel C, 32 pg; gel D, 16 pg; gel E, 6.5 pg) were electrofocused for 6.5 h and stained for protein by the method of Vesterberg (62). Parallel gels without protein were run simultaneously, cut into I-mm slices, extracted in 0.5 ml of 0.025 M KCI, and the pH of the eluates measured. The pH span of the stained protein was 6 to 8 for gel B and 6 to 7 for the remaining gels. The pH 6 boundary is at the lower end of the staining pattern. Gel F shows the pattern obtained after isoelectric focusing in 8 M urea.
Fraction VI1 (65 pg) was dialyzed under Nz overnight at room temperature against 8 M urea, 0.1 mM EDTA, and 0.1 mM dithiothreitol. Electrofocusing was for 6 h a t 6 "C (see "Methods"). The gel was stained for protein as described above. The pH values corresponding to the major protein bands are shown at the right. Gels are cut at both origin and bottom. resis in SDS corresponded to the monomer, dimer, and trimer of the 90,000-dalton subunit (Fig. 3). The trimer was the predominant species when cross-linking of Form I enzyme was done in the absence of primer. When the cross-linking was  The reaction was allowed to proceed for 90 min a t 30 "C, after which it was stopped by addition of 6 p1 of 10% SDS, 1 pl of pmercaptoethanol, and 3 min of boiling. The denatured samples were then incubated at 37 "C for 3 h. Samples of 20 pg were applied to 3.5% polyacrylamide gels and subjected to electrophoresis in SDScontaining buffer as described by Davies and Stark (25). Gels 2 and 2, primer-independent polynucleotide phosphorylase (I); gels 3 and 4, trypsin-treated enzyme (2') after incubation for 10 min in the presence of 0.14 M P-mercaptoethanol(55);gels 5 and 6, trypsin-treated enzyme after incubation for 10 min in the presence of 2 mM N-ethylmaleimide (55). Molecular weight standards are shown on the left. by guest on March 25, 2020 http://www.jbc.org/ Downloaded from performed in the presence of primer, which has been shown to interact with the primer-independent enzyme (ll), a reduction of the amount of trimeric species was consistently observed together with a concomitant increase in monomeric species.
The enzyme obtained by limited proteolysis with trypsin also showed three components after cross-linking corresponding to the monomer, dimer, and trimer of this polypeptide. In this case, however, the monomeric species was predominant, and the presence of primer during cross-linking did not affect the pattern. Treatment with thiol or N-ethylmaleimide which affects the de nouo polymerization properties of the partially proteolyzed enzyme (2), also had no effect on the cross-linking pattern (Fig. 3).
Amino Acid Composition-The amino acid composition of polynucleotide phosphorylase is shown in Table I. With the exception of the tryptophan content, the values agree with those previously published (8). Minor differences are probably due to proteolysis in the original procedure as evidenced by the previously reported heterogeneity of subunit size (8). The tryptophan content of the protein was determined both indirectly as described by Edelhoch  (Table I). The values of tyrosine and tryptophan content are in good agreement with the UV-absorption spectrum of the protein in 6 M guanidine-C1 and that of a mixture of the N-acetyl-tyrosinamide, N-acetyl-tryptophanamide, and N-acetyl-phenylalaninamide in amounts equimolar to that present in the protein (42). With the exception of a small red shift (data not shown), the correspondance of the two spectra between 260 and 300 nm was excellent.
End Group Analysis--In contrast to the E. coli enzyme, neither the primer-independent form of M. hteus polynucleotide phosphorylase nor the primer-dependent form of the enzyme obtained by limited proteolysis with trypsin contain a detectable free amino end group (Table 11). This suggests that proteolysis removes a peptide at the carboxyl end of the molecule. It is also another indication of the homogeneity of the product and low level of proteolysis that characterizes this purification procedure.
Spectroscopic Properties-The electrophoretic heterogeneity mentioned above, as opposed to the apparently homogeneous subunit composition, could be explained if oligonucleotides were tightly bound to the enzyme. The UV-absorp- tion spectrum of polynucleotide phosphorylase is shown in Fig. 4. An extinction coefficient of& " , , , = 4.80 was calculated when the protein concentration was measured by the method of Lowry et al. (43) using bovine serum albumin as a standard. This value is similar to that previously reported (2). The ratio of absorbancies at 280 and 260 nm of 1.75 indicates a very low content of nucleic acid. Contamination with 1 mol of adenosine/mol of protein subunit should increase the absorbance at 260 nm of a 1 mg/ml solution of enzyme by 0.167 and would give a 280/260 ratio of 1.08. The absence of contaminating nucleic acid was confirmed by measurement of the UV-absorption spectrum of the enzyme in the presence of 6 M guanidine hydrochloride before and after gel filtration on Sephadex G-25 to remove small size UV-absorbing material. As shown in Fig. 4, the absorption of the enzyme at 260 nm    was not significantly decreased by this treatment, indicating the absence of bound oligonucleotide. The UV-difference spectrum shown in the inset of Fig. 4 suggests that 6 tyrosyl residues and maybe 1 tryptophanyl residue are buried in the interior of the protein and become exposed upon treatment with guanidine (42). The remaining tyrosyl and tryptophanyl residues are exposed and can be detected by treatment with cetyltrimethylammonium bromide (44) (data not shown).

Oligomeric Forms of Polynucleotide Phosphorylase in Less Purified Fractions
In contrast to the purified enzyme which is greater than 90% of one size class, enzyme in crude extracts exhibited a high degree of electrophoretic heterogeneity.:' The multiple components were investigated by combining gel electrophoresis under nondenaturing conditions with a specific in situ assay for enzymatic activity (see "Methods"). Coupling this assay method with Ferguson plot analyses has enabled us to determine the number of active components present and to obtain estimates of their size and charge as a function of the degree of purification.
Three fractions of widely different degrees of purity were analyzed by this method: the crude extract (Fraction I), the Sepharose 6B fraction: and the hydroxylapatite fraction (Fraction VII) as described in the miniprint section. Table I11 '' The one-to-one correspondence between activity and protein patterns of the purified enzyme on disc gels (demonstrated in Fig. 3 of Ref. 11) could not be studied in crude fractions. However, a comparison of RF values from activity gels of fractions as a function of purification showed that the same species were present (although in different amounts) throughout the purification procedure. In addition, sedimentation equilibrium analysis of a relatively pure Sepharose 6B fraction confirmed the presence of the polymeric species seen on activity gels. The particular Sepharose 6B fraction used in this study emerged early from the column before the bulk of both the activity and the protein so that in spite of its being derived from an early stage in the purification procedure, it had only two minor contaminants visible on SDS gels (data not shown).  ellipses" are horizontally displaced against one another) (Fig.  6, lower panel). Furthermore, Set I varies in size and charge a t different stages of purification (Fig. 6, upper panel) (Table  111). Set I in the crude extract is the largest and the most highly charged. Set I of the purified species is the smallest. The Sepharose 6R fraction' (selected for its large M , ) is more heterogeneous. It contains not only a Set I similar in size and charge to that of purified enz-yme, but a Set I1 present in crude extract and an additional set of larger M, (Set 111).
Considering the assumptions in the calculation of the M, values (23), the molecular size of Set I measured by Ferguson plot analysis (210,000) and that measured by sedimentation equilibrium (237,000) are in good agreement.
Partial resolution of these sets could be obtained by centrifugation in a 0 to 20% neutral sucrose gradient. Although the enzyme activity appeared as a single peak, aliquots across this peak analyzed by gel electrophoresis showed that the relative amount of each set varied as a function of the sedimentation rate. Slowly sedimenting fractions were essentially composed of Set I, while rapidly sedimenting fractions were enriched in Sets I1 and I11 (data not shown). The enzyme thus appears to exist as a mixture of multimers which can be partially resolved by sedimentation and are, therefore, not in rapid equilibrium.
In order to study furLim the multimeric structure of the enzyme, the Sepharose 6B fraction" was subjected to electrophoresis under nondenaturing conditions. The gel was sliced, the enzyme eluted, and the fractions assayed for activity. Phosphorolysis and pol-ymerization assays of sequential gel fractions gave three peaks (Sets I to 111) with RF values identical to those seen when the whole gel was assayed directly by staining (Fig. 5, upper panel). When aliquots of fractions corresponding to Set I were then reelectrophoresed, the patterns generated were similar to those of the parent set and had bands with RF values which corresponded precisely to those of Set I in the parental species (Fig. 5, lowerpanel). On the other hand, aliquots from Set 11, when reelectrophoresed, exhibited both Set I and Set I1 activity bands, and again in the parental ratio of 65:35 (Set 1:Set 11) as determined by densitometer analysis. The RF. values were identical in each case to those of parental enzyme, although there was some enrichment in trailing or leading species as a function of the origin of the slice. Attempts to reverse the process and to generate Set I1 from Set I using purified enzyme were unsuc- [14C]poly(A) on 5% acrylamide gels. Gels were cut into 2-mm electrophoresed on a 5 9 gel; B, a polynucleotide phosphorylase slices, solubilized, and counted. Left, ["C]poly(A) synthesized by reaction mixture incubated 60 min at 0 "C before gel electrophoresis polynucleotide phosphorylase during a 2-min incubation at 37 "C was (see "Methods"). After electrophoresis. both gels were incubated in subjected to electrophoresis as described under "Methods" the presence of ADP and Mi'' and stained with acridine orange. The cessful, even when the enzyme was concentrated up to 8.4 m g / d prior to electrophoresis.

The Poly(A) Product of Polynucleotide Phosphorylase
When poly(A) is synthesized from ADP by polynucleotide phosphorylase under the conditions described under "Methods," and the reaction mixture immediately subjected to gel electrophoresis under nondenaturing conditions, two sharp acridine orange staining bands are seen in the separating gel in addition to some free enzyme at its usual RF (Fig. 7, gel B ) .
This gel was stained for polynucleotide after incubating it with ADP. Thus, any poly(A) formed prior to electrophoresis will be stained, as well as any enzyme component on the gel that can generate poly(A) during the incubation of the gel after electrophoresis. The two bands formed are characterized by very reproducible RF values (Table IV), with a coefficient of variation of 5%. The relative amounts of each band vary with the temperature and duration of incubation. Table IV, (column a) shows that Band 1 is predominant a t early times (2 to 5 min), especially when the incubation is done a t 0 "C (data not shown). However, after 7 min a t 37 "C or 60 min at 0 "C, Band 2 predominates. If the gels are stained only for poly(A) synthesized prior to electrophoresis by omitting the ADP prior to staining, then Band 2 predominates at 2 min and by 20 min is essentially the only species present (Table  IV, column b).
These sharply defined species are contrasted with the heterogeneous electrophoretic pattern of commercial poly (A). Both the staining pattern and the distribution of radioactivity of the commercial preparation show a high degree of polydispersity (Fig. 7, gel A ) . In contrast, poly(A) or [I4C]poly(A) synthesized by the purified enzyme, as described in this study, contain only 1 to 2 components, both when examined by staining or by radioactivity measurement (Fig. 7).
T o get a clearer idea of the composition of the acridine orange-stained components observed on polyacrylamide gels, we compared gel patterns of reaction mixtures with and without proteinase K treatment before electrophoresis. The gels, for convenience, have been aligned at the origin which, because of the differences in staining procedure, has caused an apparent lack of correspondance between protein and activity patterns (Fig. 8, gels A and B); the RF values are identical, however (1 1). At zero time, in the absence of proteinase K treatment, most of the enzyme is observed at the position of free enzyme (Fig. 8, gels A and B ) . However, some poly (A) is already detectable at the top of the stacking gel,

Formation of poly(A) .enzyme (band I ) and poly(A) (band 2) as a function of duration of incubation.
Aliquots from reaction mixtures at 37 "C were diluted at the times indicated and run on 5% polyacrylamide gels (see "Methods"). Gels were then incubated either in the presence of ADP and Mg2+ before acridine orange staining to visualize both enzyme activity and preformed poly(A) (column a) or in buffer alone to visualize preformed poly(A) only (column b). indicating that a small amount of polymerization had already occurred within the few seconds required for pipetting, even at 0 "C (gel B ) . However, after 2 min at 37 "C in a gel stained for protein (Fig. 8, gel C), almost all the enzyme is at the top, unable to enter the stacking gel. A gel incubated for activity (gel D) shows that, in addition, there are two acridine orange staining bands in the resolving gel. A gel incubated to detect only poly(A) synthesized before electrophoresis (gel E ) shows the usual two bands in the resolving gel with a marked predominance of Band 2. When the sample was treated with proteinase K prior to electrophoresis (gel F ) , only one of the three components seen on gel D remains, and that is Band 2 as indicated by its RF value. Thus, Band 2 is the free poly (A) synthesized by polynucleotide phosphorylase. The material which does not enter the stacking gel and Band 1 presumably are different forms of the poly(A) .enzyme complex. The fact that no new bands of poly(A) are released from the enzyme. poly (A) complexes indicates that essentially only one size class of poly(A) exists.
Proteinase K treatment confers stability to the poly(A) product by eliminating the polynucleotide phosphorylase catalyzed "transnucleotidation" reactions, which can result in a variety of sizes of poly (A) (45, 46). These reactions cause a breakdown of the single poly(A) component after storage and freeze-thawing of untreated reaction mixtures, to give a pattern somewhat similar to that of the commercial poly(A) seen in Fig. 7A. The proteinase K-treated sample, however, showed no evidence of breakdown after similar treatment (data not shown).

Size Determination of Poly(A) Synthesized de
Nouo-Proteinase K-treated and untreated aliquots of a polymerization mixture were sedimented on isokinetic sucrose gradients (47). The proteinase K-treated material moved on the gradient coincident with the 18 S marker RNA while the untreated polynucleotide phosphorylase. poly (A) mixture sedimented with a value of 23 S (data not shown). When examined on a 5 M urea-0.8% agarose gel, proteinase K-treated poly(A) migrated coincidentally with 28 S rRNA precursor which is 8,000 bases long (48). Proteinase K-treated poly(A), prepared for electron microscopy by the cytochrome-formamide Klein-Schmidt technique (49), gave a mean value of 13,000 f 2,000 nucleotides, using @X174 as an internal standard (Fig. 9).

DISCUSSION
The purification procedure described here provides significant improvements in yield and integrit5 7f structure of polynucleotide phosphorylase from M. luteus compared with previously published methods. The overall procedure, as well as individual steps, have been repeated several times and are very reproducible in terms of yield (-20%), pattern on gel electrophoresis, primer-dependence (1.1-to 2-fold), and specific activity (50 phosphorolysis units/mg).
The subunit of the primer-independent enzyme is similar in sue to the u subunit of RNA polymerase when they are electrophoresed together on SDS gels (11). Similar observations have also been made concerning the subunit of polynucleotide phosphorylase from E. coli (50,51). Although the M , of the u subunit has not been definitively established (40, 52, 53), the best fit for the M , of polynucleotide phosphorylase using other M , markers was 92,000, in agreement with the most recent value for the u subunit (40).
The cross-linking experiments indicate clearly that the enzyme is trimeric, as has been shown to be the case for the E.
coli enzyme (54). The cross-linked trimer of the independent forms of bolh enzymes is easily obtained, whereas the primerdependent form (Form T) of M. luteus polynucleotide phosphorylase is more resistant to cross-linking. Although treating this latter form with a sulfhydryl reagent restores its primer independence in terms of its enzyme activity (2, 55), it does not change the cross-linking pattern. This indicates that the groups which react with dimethylsuberimidate are either eliminated during the trypsin treatment, or that there has been a change in subunit orientation. Treatment of the primer-independent form with (AP)~A changes the structure so that fewer cross-linked trimers are formed and more monomer remains. Previously, a relatively large conformational change induced by oligonucleotide binding was deduced from the dramatic change in gel electrophoretic pattern of primer-independent enzyme after incubation with oligonucleotides (11). On the other hand, no changes in gel patterns were seen with Form T enzyme when it was incubated with oligonucleotide,5 in agreement with the lack of effect of oligonucleotides on the cross-linking patterns. This trimeric structure does not appear to be the only form of the enz-,me in crude extracts since at early stages of purificatior., significant amounts (30 to 40%) of higher molecular weight components (Sets I1 and 111) were always observed as indicated by Ferguson plots and sucrose gradient centrifugation. The cause of this association is not yet clear. Since we were unable to generate these forms with more purified preparations, it is possible that an oligonucleotide or an unidentified protein is required.
The difference in Set I M , between the more purified fractions (220,000 and 210,000) and the crude filtrate (320,000) is especially interesting in view of the report that the p form of the E. coli enzyme has a subunit structure a&, and a M , of 365,000 (56). It is possible that a similar heteropolymer exists in M. luteus, but that the p subunit is lost during purification. I t is also possible that oligo or polynucleotides are bound to the M. luteus enzyme in the crude extract and are contributing to the sue, since (Ap),A changes the gel pattern of the purified enzyme back to that found in the least purified fractions (1 1). The aspn form of the E. coli enzyme has a 280/260 ratio of 1.0, indicating that it does contain bound nucleic acid (56).
The amino acid composition of the E. coli (Table I).
It has been suggested that tryptophanyl and tyrosyl residues in oligopeptides serve to anchor single-stranded nucleic acids to proteins (58). Thus, it is possible that the exposed tryptophan, along with the 6 exposed tyrosyl residues of the primerindependent enzyme, allows ADP binding by stacking type interactions and thus de novo polymerization. The loss of 1 tryptophan and 3 tyrosyl residues in the conversion of Form I to T, may explain why Form T has lost both the ability to bind oligonucleotides with high affinity and to catalyze de novo synthesis of poly (A).
De Novo Poly(AJ Synthesis-Under the conditions used (5 PM enzyme, 25 mM ADP, and short incubation times), polynucleotide phosphorylase synthesizes a surprisingly narrow size class of very long chain poly(A) molecules. These can be readily isolated following proteinase K treatment of reaction mixtures, thus allowing for the rapid preparation of a large, relatively homogenous, protein-free, polynucleotide. This poly(A) is characterized by a very sharp band on nondenaturing polyacrylamide gels in contrast to commercial poly (A) preparations which display a wide range of sizes under these same conditions.
The three poly(A) species seen when untreated polynucleotide phosphorylase reaction mixtures are electrophoresed are most likely formed according to the following reaction scheme: The clear separation of these forms was not seen in the previous studies analyzing poly (A) on gels (4, 9), in part because samples from reaction mixtures were treated with EDTA to stop the reaction, but were not deproteinized prior to electrophoresis. Under these conditions, we also find much of the material concentrated at the interface of the separating and stacking gels (data not shown). Apparently, EDTA partially dissociates the network of poly (A), . enzyme,,,, such that a smaller network is produced which now can move through the stacking gel, although it still cannot enter the resolving gel. Since a poly(A) I enzyme complex can be visualized separately as Band l, it seems likely that it is a more complex matrix-type structure which remains at the top of the stacking gel. Harvey et al. (10) had previously postulated a poly (A). enzyme network to explain the very large increase in viscosity which accompanies early stages of polymerization. They suggested that poly(A) chains attached to one enzyme molecule might subsequently become attached to other enzyme molecules to yield a three-dimensional, cross-linked network. Several observations from their work and that of others support the idea of such a matrix: 1) no oligonucleotides are released during synthesis, ie. a processive reaction occurs in which poly (A) is not released until synthesis is complete so that fulllength poly(A) remains bound to the enzyme, and 2 ) the stability of the poly (A) .enzyme complex is high, as evidenced by its isolation on electrophoresis and its adsorption to Millipore filters (1, 10, 59).
It is difficult to obtain a precise size for the poly(A) synthesized by M. luteus polynucleotide phosphorylase. Although 5 M urea-0.8% agarose gels yield a value of 8,000 bases, electron microscopy shows a distribution with a mean of 13,000 f 2,000 bases. Since the urea gels may not be completely denaturing for poly (A), it seems likely that the electron microscopy may be closer to the true size. A high weight average degree of polymerization was predicted by Peller and Barnett (60) to be an intermediate stage in the unprimed synthesis of poly(A) as a result of the more favorable kinetics for adding a residue in contrast to reinitiating another polymer chain. However, it remains to be found by what mechanism polynucleotide phosphorylase is able to terminate synthesis in a pseudo-synchronous fashion yielding a very homogenous polynucleotide. 14.