RNA Polymerase from Clostridium acidi-urici CHARACTERIZATION OF A NATURALLY OCCURRING RIFAMPICIN-RESISTANT BACTERIAL ENZYME*

We report here the isolation of a prokaryotic RNA polymerase that shows pronounced template specificity. The enzyme from Clostridium acidi-urici is highly active on DNA templates isolated from phage that infect Gram-positive organisms and is essentially inactive at either high or low ionic strength on DNA from phage associated with Gram-negative bacteria. The enzyme is also unique among RNA polymerases isolated from wild type bacteria in being highly resistant to inhibition by rifampicin. These properties are characteristic of the enzyme present in several independently isolated strains of C. acidi-urici However, RNA polymerase present in other clostridial species resembles the enzyme present in Bacillus subtilis in sensitivity to rifampicin and template specificity.

We report here the isolation of a prokaryotic RNA polymerase that shows pronounced template specificity. The enzyme from Clostridium acidi-urici is highly active on DNA templates isolated from phage that infect Gram-positive organisms and is essentially inactive at either high or low ionic strength on DNA from phage associated with Gram-negative bacteria. The enzyme is also unique among RNA polymerases isolated from wild type bacteria in being highly resistant to inhibition by rifampicin. These properties are characteristic of the enzyme present in several independently isolated strains of C. acidi-urici However, RNA polymerase present in other clostridial species resembles the enzyme present in Bacillus subtilis in sensitivity to rifampicin and template specificity.
Analysis of the base sequence data of the 16 S rRNA of the prokaryotes has served as the basis for the reclassification of these organisms into two kingdoms (1). Most bacteria belong to the eubacterial kingdom (2), but it has been suggested that methanogens, extreme halophiles, and various thermoacidophiles comprise a separate line of descent among the prokaryotes to be called the Archaebacteria (3). This reclassification is consistent with relationships based on comparative analyses of protein sequences and cell wall composition and suggests relationships not previously recognized. Thus, most of the Gram-negative nonphotosynthetic organisms, including the enterics, are classified in a subline of a major eubacterial group, the purple bacteria, while Gram-positive organisms, including the clostridia, bacillae, and lactobacillae, occur in another major group, the "clostridia." The relatively distant relationship between Gram-negative and Gram-positive bacteria suggested by the comparative analysis of the 16 S rRNA sequences was of particular interest to us in relationship to the observation made in this laboratory (4) that ribosomal protein-synthesizing systems derived from Gram-positive organisms were unable to translate mRNAs derived from Gram-negative organisms or phages related to these organisms. It was subsequently reported that genes from Gram-negative organisms carried on hybrid plasmids were not expressed in vivo (5) by Gram-positive organisms. These observations both confirmed the earlier in vitro translational studies and raised the question of the possibility of transcriptional barriers to heterospecific gene expression.
RNA polymerase (EC 2.7.7.6) has been purified from relatively few Gram-positive bacterial species and not from any clostridial species (6). The subunit composition of the normal vegetative enzyme from Gram-positive species that have been * 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. examined resembles that of the enzyme from Gram-negative sources except for the fact that the a subunit of the enzyme from four species of Bacillus and Lactobacillus curvatus has a molecular weight about one-half of that of the a subunit of the Escherichia coli enzyme or any of the other enzymes isolated from Gram-negative species. The enzymes that have been isolated up to now from Gram-positive sources do not differ in any significant properties from those isolated from Gram-negative bacteria with respect to specificity of promoter utilization or antibiotic sensitivity, although differences in efficiency of promoter utilization have been noted as well as susceptibility to ionic conditions (7)(8)(9). RNA polymerases from the Archaebacteria, however, may be distinguished from those derived from eubacteria on the basis of their subunit composition as well as by the rifampicin insensitivity of the enzyme (10,11).
Because the clostridia, like the methanogens of the archaebacteria, are strict anaerobes and are considered to closely resemble the first organisms in primeval evolution (2), and because of the observations described above, it appeared worthwhile to us to examine the properties of RNA polymerase from clostridia. The results obtained are described here and appear to justify our curiosity since it resulted in the discovery of an RNA polymerase with unique properties among those of bacterial origin.

MATERIALS AND METHODS
The sources of most materials, including bacteriophage DNAs, and methods have been previously cited (9). X4, a Myxococcus phage, was obtained from David R. Zusman of the Department of Microbiology and Immunology of this University. Other materials were obtained as follows. Rifampicin was obtained from Sigma; Bio-Gel A-1.5m (100-200 mesh) from Bio-Rad; uric acid from Pfanstiehl; and poly(dA-dT) (alternating polymer) from Miles. Polymin P (polyethyleneimine) was purchased from Eastman and a 10% (v/v) stock solution was prepared as described by Burgess (12). The (20,21) followed by autoradiography. B. subtilis RNA polymerase transcripts of 4)29 DNA (9) were used as standards. For analysis of proteins, polyacrylamide gel electrophoresis in the presence of SDS was performed in slab gels (22) with a discontinuous buffer system (23) and an acrylamide/ bisacrylamide ratio of 30:0.8.
Calculation of Subunit Molecular Weights-Molecular weights of subunits were determined on 7.5% SDS-polyacrylamide slab gels. The /3' (Mr = 160,000), /3(150,000), and a (36,500) subunits of E. coli RNA polymerase (24,25), the a (Mr = 56,000) and a (43,000) subunits of B. subtilis RNA polymerase (6), and C. acidi-urici formyltetrahydrofolate synthetase (monomer M, = 60,000) (26) were used as standards. Because of the large variation in reported values for the E. coli a subunit and the B. subtilis /8 and /3'subunits, the molecular weights were calculated from migration distances in our buffer system. Purification Procedure-RNA polymerases from C. acidi-urici (strain 9a) and C. cylindrosporum (strain HC-1) were purified by a modification of the procedure of Burgess and Jendrisak (12). All steps were carried out at 4 'C.
Frozen cells (25 g) were suspended in 75 ml of Buffer I and sonicated for 1 min with a Branson Sonifier (model J-17A). The crude extract was then diluted with 100 ml of Buffer I and centrifuged at 15,000 rpm in the SS-34 rotor of a Sorvall RC-5B centrifuge for 30 min. The supernatant solution was decanted and a 10% (v/v) solution of Polymin P was added dropwise with stirring to a final concentration of 0.2%. After stirring 10 min, the suspension was centrifuged 15 min at 6000 rpm and the clear supernatant solution was discarded. The pellet was resuspended with the aid of a Dounce homogenizer in 100 ml of Buffer II containing 0.5 M NaCl. The suspension was stirred for 10 min, centrifuged as before, and the supernatant fraction was discarded. The pellet was then extracted with 80 ml of Buffer II + 1.0 M NaCl. After stirring for 10 min, the mixture was centrifuged for 30 min at 6000 rpm and the precipitate was discarded.
The RNA polymerase in the supernatant solution was then precipitated by the addition of solid ammonium sulfate to a final concentration of 65% (39.8 g/100 ml), with the addition of 5 N NH40H to maintain pH 7.5. The suspension was stirred for 30 min and centrifuged for 30 min at 18,000 rpm. The pellet was dissolved in a minimum volume of Buffer III (VT = 3 ml) and applied to a column (1.2 x 68 cm) of agarose A-1.5m equilibrated with the same buffer. Fractions containing polymerase activity were pooled and diluted with 20 mM Tris-HCl (pH 8), 0.1 mM EDTA, 0.3 mM dithiothreitol, and 20% glycerol until the conductivity was equal to that of Buffer IV.
The diluted enzyme (50 ml) was applied to a column (1.2 x 7 cm) of DNA cellulose equilibrated with Buffer IV and washed with 10 column volumes of buffer. The polymerase was then step eluted with Buffer IV + 1.0 M NaCl. The enzyme was stored in this buffer at 4 'C and was stable for at least 18 months. A summary of the purification is given in Table I. Purification of the enzyme from C. 'The abbreviation used is: SDS, sodium dodecyl sulfate. cylindrosporum (strain HC-1) was carried out by the same procedure and gave a similar yield of enzyme of the same specific activity.
Partial Purification of RNA Polymerases from Other Clostridial Strains-Enzymes from C. pasteurianum, C. acidi-urici (strains AAM-2 and AC-1), and C. cylindrosporum (strain MJ-6) were purified from 10 g of cells as outlined above through ammonium sulfate precipitation. The activity in the resuspended pellets (5 ml) was entirely DNA-dependent and proportional to the amount of enzyme added.

RESULTS
Subunit Composition-The subunit structure and purity of two preparations of RNA polymerase from C. acidi-urici are shown in Fig. 1 (lanes 2 and 3). For comparison, the enzymes purified from B. subtilis (lane 1) and E. coli (lane 4) are shown. The enzyme obtained from C. acidi-urici after step 4 of the purification (lane 3) shows six components. If these are compared with the component subunits of RNA polymerase from E. coli, starting with the high molecular weight components and going to the lower molecular weight bands, one can identify the two largest components of the C. acidi-urici enzyme as the ,B and,' subunits. We are designating the larger subunit ,8 and the smaller one /B' by analogy to the findings reported for the other enzymes purified from Gram-positive sources (27). We have observed that the smaller of the two subunits of the C. acidi-urici enzyme binds to DNA, a characteristic of the /3' subunits (28). The next smaller component has a molecular weight very similar to that of the a subunit of the E. coli enzyme. However, we do not believe that this is the normal a subunit of the C. acidi-urici enzyme since it is removed following step 5 in the purification to yield an enzyme (lane 2) with even higher specific activity and no change in template specificity. The a subunit of the C. acidi-urici enzyme appears to be the next component which is retained in the preparation even after step 5 of the purification and has a molecular weight similar to that of the a subunit of the B. subtilis enzyme (lane 1). Finally, the a subunits of this enzyme, as well as those of the other bacterial polymerases examined here, appear as doublets. This behavior is not uncommon (29).
The molecular weights of the subunits of these enzymes are summarized in Table II. The molecular weights of the a subunits are within the range of those reported for other Gram-positive bacteria (6). The /3 and /' subunits of the Purification of RNA polymerase from C. acidi-urici (strain 9a) Starting material was 25 g wet weight of C. acidi-urici. Activity was assayed with 429 DNA as described under "Materials and Methods" with 1 unit of activity representing 1 nmol of cmp incorporated in 10 min at 37 'C. Fraction 5 polymerase is the result of applying the fraction of enzyme that passed through DNA-cellulose column 1 to a second column and eluting it with buffer containing 1 M NaCl. It is unique in being free of 95,000 component (see text). The total recovery of activity in this preparation (fractions 4 and 5) is 81%. Other preparations were usually only carried through step 4 and resulted in recoveries of 80% with specific activities of 2500.  Ftc. 1. Subunit composition of C. acidi-urici RNA polymerase. Samples were analyzed by electrophoresis in a p.5 l, pol acrxIamide slab gel as described under -Materials and Methods. Lane 1. 6 pg of B. subtilis holoenzvne (a saturated). Lane 2, 8 pg of C. cidiuirici enz-me after step S of the purification. Lane 3, 6 ag of C. acidiurici enzvme after step 4 of the purification. This enzxrne is estimated to be 25-30%' a saturated. Lane 4, 6 pg of F. coli holoenzvnme (a saturated). The cx subunits of all of the bacterial enizvmes usuallx run as doublets in the gel systenm. It is; assumed that the (Uacic. ur enzxme has two identical a subuniits. enzyme from all six of the clostridial strains that we have examined have identical mobilities on polyacrylamide gels (data not shown).
The RNA polymerase prepared from C. aucdi-arien by the procedure described here, as well as by the method of Duffy and Geiduschek (30), consistently co-purified with a protein of M, = 95,000 (the third protein component in lane 3, Fig. 1).
This protein may be analogous to proteins of similar size that are present in preparations of RNA polymerase from a variety of bacterial species (25,27,(31)(32)(33)(34). Values reported for the molecular weights of these proteins varv from 84,000-110,000. In all cases, the migration of the protein in SDS-polyacrvlamide gels closely resembles that of the E. coli a subunit. It has been found in other systems (25,(31)(32)(33) that this protein can be separated from polymerase either by zone centrifugation through a glycerol gradient or by chromatography on single-stranded DNA agarose. However, the clostridial en-zyme was found to havxe a very loosely associated a subunit so that both chromatography on the single-stranded DNA analogue heparin agarose and zone centrifugation cause dissociation of the enzyme into the core and a subunits. We have been able to obtain a preparation of enzyme free of this protein onlv in the following manner. The DNA cellulose column employed in the final step of the purification was overloaded during one enzyme preparation. Although there was no RNA polymerase activitv (or /A/3' material) in the early passthrough fractions, substantial amounts of polvnmerase were present in later fractions. Examination of these fractions on a polyacrylamide slab gel revealed that they contained none of the 95,000 component, all of which had bound to the I)NA column. The pass-through fractions containing polymerase activity were then pooled and applied to a second column of DNA cellulose. The enzvme that eluted with 1 M NaCl was completely free of the 95,000 component. The presence of this protein does not appear to alter any of the properties of the polymerase, and we have used the enzyme obtained after step 4 of the purification for some of our studies. A summarv of' the purification is given in Table I. It should be noted that the total recovery of activitv is roughl 80(. Resistance to Antibiotics Sensitivit' to the antibiotic rifampicin is a general characteristic of RNA polvmerases of eubacteria. Under assav conditions employed in these studies, the B. subtills enzvme is inhibited 9t5 by 0.7 gg/ml of rifampicin ( Fig. 2), which is tvpical of reports that 0.02 to I gg/ml of rifampicin causes conmplete inhibition of activity of the enzvme from bacterial sources (35)(36)(37). As shown in this figure, the enzyme from C. acidi-urici is extremely resistant to inhibition by this antibiotic. At the highest concentration of rifampicin tested, 400 pg/ml, the enzyme still shows 30's activity at low ionic strength.
Streptolydigin, an antibiotic that has been found to inhibit phosphodiester bond formation (elongation) in the transcription process (38), was found to completely inhibit the clostridial enzyme at the lowest level tested, 10`m.
Template Specificity Three templates were used to quantitate the recovery of activity during the purification of the C.
acidi-urici enzyme: B. subtilis phage <29 DNA, F. coli phage T7 DNA, and poly(dA-dT). It was found at every stage of the purification that the enzyme was only 1-5% as active on T7 DNA as on ¾29 DNA. The enzyme was highly active on poly(dA-dT). This apparent I)NA template specificity is unique among bacterial RNA polymerases that have been described (8) and led us to determine the activitv of this  Table III. TFhe specific actixity of the enzyme xaried over a 100-fold range as a function of the template on which it xwas assayed. In general, much higher levels of activ>itv xvere found on DNAs from bacteriophages that infect Gram-positiv.e organisms. Consistently loxw levels of activitv were found on DNAs from Gram-negative sources. The effect of salt on transcription will be discussed below.
Char-acteri&atiotn of 629 Transcr4pts-The in ritr o transcription prolucts made by B. sarbtilrs RNA poly-merase in response to @29 I)NA have been well characterized (39). It was, therefore, possible to determine xw hether the same promoter and termination sites are utilized by the C. ciclciarlci enzyvme as bv the B. sahtilis enzymie. As shown in Fig. 3  Characterization of Interaction with T7 DiNA Although the C. acudi-urici RNA polymerase shows a very low level of activity on TY DNA, some products can be detected (Fig. 4, lanes d and f). These products resemble those formed bv the B. saibtlli's enzvme on 1T? (Fig. 4, latne b) rather than those formed bv the E. coli enzyme (Fig. 4, lane a) formiation of significant amounts of C and I) transcripts in addition to the A transcript, the main product synthesized by the E. coli enzyme. The A transcript is a mixture of products initiated at three strong A promoters that are located very near the end of the T7 genome (40). These products are similar in size and are not separated in the gel system used here. In order to determine whether the C. acidi-u-iaci enzyme utilizes any of the three A promoters of T 7 DNA or simply initiates at the end of the DNA to form the smiall amount of A-like product observed, transcription of the T7 deletion mutant AD11 hby the C. aucdi-aricl enzyme was determined. This mutation deletes the region betwxeen 1.3`and 4.1`%' of the genome that includes the A> and A1 promoters and only the A, promoter is present (40). Only 30'% as much acidinsoluble radioactivity was incorporated by the enzyme in response to the mutant DNA compared to the parent Tl DNA. Examination of the RNA svnthesized from AD1 1 (Fig.  4, lane g) showed a single faint band corresponding in size to the C transcript. Since the efficiency of end initiation should be the same for both wild type T7 and the deletion mutant, it appears that the 8000 nucoeotide transcript from wild tvpe T7 DNA is a result of initiation at the A, and/or A: promoter. The absence of anv higher molecular weight RNA indicates that the enzyme terminates transcription efficiently at the t8.9%. terminator.
In order to substantiate this conclusion and to further characterize the site of interaction of the C. acidi-urici enzxvme with T7 DNA, a Hae III digest of T7 DNA was transcribed.
This endonuclease cleaves FT7 DNA at a position 1450 base pairs from the left end of the T7 genome and transcription of the restricted DNA by F. coli RNA polvmer-ase vields three RNA species that are easily separated by electrophoresis on a 2.5% agarose gel (Fig. 5, lane i). TIlhe three transcripts containing 948, 828, and 712 nucleotides result from initiation at the A1, A, and A} promoters, respectively. As shown in lane g of Fig. 5, the C. au.ldi-urici enzyme yields a single transcript that coincides in size with that fornmed from the A, prornoter. The efficiencv of transcription from this promoter is low. In order to synthesize enough labeled JRNA to be visible on the gel, it was necessary to double the concentrations of enzyme and DNA and to increase the specific activity of the [x-2P]ATP relative to that used for the E. coli enzvme.
Salt Effect Since ionic strength affects every step in the transcription process (41), we have determined the effect of ionic strength on transcription of p29 DNA by C. acicli-ia-icni RNA polvmerase as measured by the overall reaction (Fig. 6). Optimal activity is observed at 160 mm KCl, the samne salt concentration at which the B. sabtilis enzvymie exhibits maximum activity (9). However, the extent of stimulation by salt is much higher for the clostridial enzyme (2.5x) than for the B. subtilis enzvme (1.3x). A comparison of transcripts produced at low and high ionic strength (Fig. 3 amounts at low ionic strength (GI and A1) terminate at the end of the DNA (39). As shown in Table III, transcription of 429, 4e, or SPOI DNA by the clostridial enzyme is very efficient in the presence of 160 mM KCl. In contrast, the low level of activity on the templates related to Gram-negative organisms is essentially eliminated by high salt. At low ionic strength, the specific activity of the purified clostridial enzyme is slightly higher on poly(dA-dT) than on 429 DNA, while at high salt, the enzyme is only 37% as active on the synthetic template as on 429 DNA.
Other Clostridial Enzymes-The unusual properties of the enzyme from C. acidi-urici (strain 9a) prompted us to investigate whether these properties were characteristic of all clostridial RNA polymerases. C. acidi-urici is classified in group V of the clostridia, those with special growth requirements (42).
As a representative of a more typical clostridial species, we chose C. pasteurianum, a member of group I of the genus Clostridium. The properties of the RNA polymerase from C. pasteurianum (Table IV) were very similar to those of the enzyme purified from B. subtilis. The C. pasteurianum enzyme was active on T7 DNA and showed the template dependent salt effect that seems to be characteristic of polymerases isolated from Gram-positive organisms. The RNA polymerase from C. pasteurianum was also found to be sensitive to low levels of rifampicin.
C. cylindrosporum is another purine-fermenting organism that, like C. acidi-urici, can be classified as a group V Clostridium. The RNA polymerase from C. cylindrosporum (strain HC1) was purified to about 95% homogeneity. The specific activity of the purified enzyme on 429 DNA was identical with that of the enzyme from C. acidi-urici and showed a similar response to increasing ionic strength. However, the RNA polymerase from C. cylindrosporum is inhibited 97% by 0.7 gg/ml of rifampicin and transcribed T7 DNA at low ionic strength 50% as efficiently as it did 4)29 DNA. The products of the C. cylindrosporum enzyme on 429 DNA (Fig. 3, lanes f and g) and on T7 DNA (Fig. 4, lane c) were identical with those formed by B. subtilis.
The difference in the effect of rifampicin on the RNA polymerases of C. cylindrosporum and C. acidi-urici was reflected very dramatically in the effect of the antibiotic on the growth of these two organisms. The effect of rifampicin on the 24-h growth of the two organisms was determined at 10-fold dilutions of the antibiotic from 200-0.00002 ,g/ml. Growth of C. cylindrosporum, strain HC1, was completely inhibited by 0.0002 jug/ml, but was unaffected by the 10-fold lower dilution tested. Growth of C. acidi-urici, strain 9a, was completely unaffected by up to 2.0 gg/ml of rifampicin, but was finally inhibited at a concentration of 20 gg/ml. Since spontaneous rifampicin mutations occur at a frequency of one in 108 (37), we wished to determine whether C. acidi-urici strain 9a was simply the chance isolation of a rifampicin-resistant spontaneous mutant or was representative of a class of soil organisms. We, therefore, partially purified RNA polymerase from two different strains of clostridia, AAM-2 and AC-1, that had previously been classified as strains of C. acidi-urici (15) on the basis of spore morphology and fermentation products (43). As shown in Table IV, the enzymes from strains AAM-2 and AC-1 showed the same resistance to inhibition by rifampicin as did strain 9a. The enzymes also transcribed T7 DNA with very low efficiency. Strain MJ-6 had previously been classified as a strain of C. cylindrosporum (15), and the RNA polymerase from this organism showed the same sensitivity to rifampicin as did the enzyme from strain HC1 of C. cylindrosporum.
Transcription with Combinations of Heterologous Subunits-Previous experiments (39) based on the transcription of )29 DNA and T7 DNA by E. coli core polymerase with either B. subtilis or E. coli 4 subunit demonstrated that the transcription products were characteristic of the core component. Since C. acidi-urici RNA polymerase shows such a marked difference in its ability to transcribe T7 DNA compared to polymerase from E. coli, it appeared worthwhile to confirm the conclusion concerning this core function.
The experiments illustrated in Fig. 5 demonstrate that the transcription products formed by bacterial RNA polymerase composed of heterologous mixtures of core and a elements are characteristic of the core component rather than of the a subunit. Since the a subunit of C. acidi-urici RNA polymerase had not been purified to homogeneity, the holoenzyme was   TABLE IV DNA specificity and rifampicin sensitivity of clostridial RNA polymerases Transcription reactions (50 gl) were carried out on 1-1l aliquots of partially purified enzymes as described under "Materials and Methods." Inhibition by rifampicin was determined in the presence of 160 mM KCl. The protein concentration in all of the redissolved 65% ammonium sulfate pellets was -5 mg/ml. The assays were incubated for 10 min at 37 'C. Activity was measured by following trichloroacetic acid-precipitable radioactivity. used as a source of the a subunit in an exchange reaction already demonstrated to occur between the E. coli core and the B. subtilis a subunit present in the B. subtilis holoenzyme (39). The E. coli core enzyme had a significant level of activity on T7 DNA as measured by trichloroacetic acid-precipitable counts (see legend to Fig. 5), but the products of transcription (lane b, Fig. 5) are different from those made by E. coli holoenzyme (lane a, Fig. 5). The core products are probably the result of initiation at nicks or at the ends of the DNA (7). The addition of purified E. coli a subunit to the core polymerase stimulates activity 10-fold, and the products (lane d) are specific and characteristic of the E. coli holoenzyme (lane a). Transcription of T7 DNA by an equivalent amount of C. acidi-urici RNA polymerase alone does not yield any detectable products (lane c) under these reaction conditions. The addition of E. coli core polymerase to a transcription reaction containing the C. acidi-urici holoenzyme results in the formation of products (lane e) characteristic of the E. coli holoenzyme. The level of activity is comparable to that found with the homologous reconstituted E. coli enzyme. The addition of purified E. coli a subunit to the C. acidi-urici polymerase (lane f) does not result in the formation of any detectable transcripts.
The role of the core components in determining the extent of promoter expression was also tested through the use of the Hae III digest of T7 DNA, since it has been demonstrated that the C. acidi-urici holoenzyme initiates transcription at the A2 promoter only (lane g, Fig. 5), while the E. colireconstituted enzyme also utilizes the A1 and A3 promoters (lane i, Fig. 5). The addition of C. acidi-urici RNA polymerase as a source of the a subunit to the E. coli core (lane h) results in the formation of transcripts of Hae III-digested T7 DNA that are characteristic of the E. coli enzyme (lane i). Transcription products from Hae Ill-digested T7 DNA with E. coli core polymerase alone are shown in lane j.
Although we have not prepared C. acidi-urici a subunit in pure form, we have obtained pure C. acidi-urici core in the course of attempting to purify this enzyme by the heparinagarose procedure (9). The enzyme in the crude extract bound to a column of heparin-agarose and was eluted with 0.5 M NaCl. Fractions containing the enzyme were then applied to a glycerol gradient in a zonal rotor. Following centrifugation, there was no 429-dependent activity in any of the fractions of the gradient. However, by assaying mixtures of various fractions, it was discovered that the enzyme had dissociated into core and a subunits. The core polymerase was near the bottom of the gradient and was essentially pure, although the a subunit present in other fractions was still impure.
Since purified a subunits of E. coli and B. subtilis were available (39), the activity of the heterologous RNA polymerase containing C. acidi-urici core subunits with a subunits derived from other sources was tested (Table V). When as- sayed on 029 DNA, C. acidi-urici core or a subunit alone was inactive. However, the C. acidi-urici core formed transcripts on 429 DNA when mixed with a subunits derived from C.
acidi-urici or B. subtilis, but not from E. coli. The transcripts made by the two active reconstituted enzymes were identical with those made by C. acidi-urici holoenzyme (data not shown). When tested on T7 DNA, none of the heterologous mixtures described above showed any activity above that of the C. acidi-urici holoenzyme.

DISCUSSION
Clostridial RNA polymerase resembles the enzyme from other Gram-positive organisms in subunit structure (,13',a,a2) and in having a a subunit roughly one-half the molecular weight of that present in E. coli or other Gram-negative organisms. This subunit structure is characteristic of the enzyme isolated from C. cylindrosporum as well as that from C. acidi-urici both classified as members of clostridial group V, species with special growth requirements (42). In this respect, these clostridial enzymes differ from those found in the Archaebacteria, which appear to have subunit structures markedly different from that found in the organisms of the eubacterial kingdom.
Although the RNA polymerase of C. acidi-urici resembles the enzymes found in other eubacteria with respect to subunit composition, it differs significantly in two other respects: its apparent specificity in template recognition and its resistance to inhibition by rifampicin. The C. acidi-urici enzyme shows very low activity on DNAs derived from phages that infect Gram-negative organisms. We do not believe that this phenomenon is attributable to the presence of a high proportion of inactive enzyme molecules as has been demonstrated to explain the relatively poor efficiency of RNA polymerase from several bacterial orders in transcribing T7 DNA compared to the activity of the E. coli enzyme on that template (8,44). This view was substantiated by the observation that the ratio of the activity of such enzymes to that of the E. coli enzyme on T7 DNA is the same as the ratio of the activities of the two enzymes on poly(dA-dT) (8). This is not true of the C. acidi-urici enzyme. Although it shows low activity on T7 DNA, it shows high activity on poly(dA-dT) as well as on DNAs derived from phages related to Gram-positive organisms. This is true whether the activity is determined at low or high salt concentrations. In addition to possessing the unique template "specificity," RNA polymerase from C. acidi-urici is also distinguished from all other RNA polymerases isolated from wild type prokaryotes in being resistant to rifampicin. In this respect, it resembles the enzyme from eukaryotes and from the Archaebacteria. These unusual properties of the enzyme were associated with several other independently isolated strains of C.
acidi-urici but not with the enzyme present in C. pasteurianum, a member of group I clostridia, or of another species of group V clostridia, C. cylindrosporum. Although C. cylindrosporum is very closely related to C. acidi-urici in its phenotypic characteristics in that both organisms can only utilize a limited number of purine bases as carbon, nitrogen, and energy sources (43), they have been shown to be genetically distantly related on the basis of several protein sequences (15). The classiflcation of Woese (2) also suggests that C. acidi-urici is somewhat distantly related to other clostridia, but it has not been possible to include C. cylindrosporum in the studies of ribosomal 16 S base sequences needed for phylogeny determinations.
The extremely low level of activity of C. acidi-urici RNA polymerase on T7 compared to that of other eubacterial enzymes has prompted us to characterize this enzyme as possessing template "specificity." However, it must be recognized that the C. acidi-urici enzyme is clearly related to the other eubacterial enzymes. Thus, although the activity of C. acidi-urici polymerase on T7 DNA is extremely low, the enzyme does recognize three of the T7 early promoters, A2, C, and D. The ability to select promoter sites is a property of the a subunit (7,32) and the C. acidi-urici a subunit appears to be homologous to other eubacterial a subunits in this respect. In addition, the C. acidi-urici a subunit interacts with the E. coli core polymerase to give a highly active heterologous enzyme. However, strikingly different amounts of products are synthesized on T7 DNA by the C. acidi-urici polymerase than by any other eubacterial enzyme tested. We believe that the results reported here related to the activity of heterologous mixtures of a subunits with core enzymes confirm those previously reported (39) and demonstrate that the yield of each transcript formed by such enzymes is characteristic of the core elements. The examples described in the current investigation are particularly striking since the transcripts of the C.
acidi-urici RNA polymerase on T7 DNA and Hae III-digested T7 DNA are so easily distinguished from those of E. coli RNA polymerase on these templates. However, these results were obtained by measuring only productive transcription and conclusions based on binding or total initiation (including abortive starts and paused products) might be somewhat different.
Although we have previously shown (39) that transcripts formed by a heterologous mixture composed of E. coli core and B. subtilis a on 429 DNA are characteristic of the core element, it was not possible to carry out the experiment with the reciprocal combination of B. subtilis core with E. coli a since this combination of subunits is inactive on any DNA template tried (39). It was assumed that steric factors prevent the interaction of these heterologous subunits since the a subunit of the E. coli enzyme is almost twice as large as that of B. subtilis. Consistent with these findings, it has also been found in the studies reported here that the heterologous enzyme composed of C. acidi-urici core with E. coli a is inactive on any DNA template tested. However, the role of the core elements in affecting the transcripts could be demonstrated with a heterologous enzyme derived from Grampositive organisms. The heterologous enzyme composed of C. acidi-urici core with B. subtilis a subunit forms transcripts on 429 DNA indistinguishable from those formed by either holoenzyme, which are indistinguishable from each other. Although this experiment demonstrates that a heterologous combintation of subunits can generate an active enzyme, it does not serve to distinguish unique functions of the subunit components of the enzyme. However, this heterologous mixture, although active on (29 DNA, exhibits extremely low activity on T7 DNA. This behavior is characteristic of the core element rather than of the a subunit of the heterologous enzyme.
In a sense, the results reported here suggest that C. acidiurici RNA polymerase represents an extreme example among the enzymes of Gram-positive bacteria with respect to apparent promoter speciflcity. This concept depends on a recognition of the effect of ionic strength on the activity of RNA polymerase. The activity of the enzyme fromE. coli, representative of Gram-negative bacteria, is relatively unaffected by ionic conditions whether transcribing DNA templates from Gram-negative or Gram-positive sources. The RNA polymerase of Gram-positive bacteria, such as B. subtilis, is affected by ionic conditions, and the effect is a function of the nature of the DNA template. The enzymes are markedly inhibited at high salt when transcribing DNA from Gram-negative sources, like T7, but are not inhibited, or may be stimulated, by high salt when transcribing DNA from Gram-positive sources such as 429 DNA. This effect is even more pronounced in the case of the RNA polymerase of C. acidi-urici, where transcription of T7 DNA at low salt concentrations is very poor, and the enzyme is virtually inactive on this DNA at high ionic strength, whereas it transcribes 429 DNA most efficiently at high ionic strength. These observations suggest that the transcriptional process may play some role in the restriction of genetic expression observed in vivo with systems derived from Gram-positive organisms.