Proc. Natl. Acad. Sci. USA Vol. 93, pp. 11580-11585, October 1996 Biochemistry

Identification of the coding region for a second poly(A) in Escherichia coli GONG-JIE CAO*, JOE POGLIANOt, AND NILIMA SARKAR*t *Boston Biomedical Research Institute, Boston, MA 02114; and Departments of tBiological Chemistry and Molecular Pharmacology and tMicrobiology and Molecular , Harvard Medical School, Boston, MA 02115 Communicated by Sankar Adhya, National Cancer Institute, Bethesda, MD, July 17, 1996 (received for review April 25, 1996)

ABSTRACT We had earlier identified the pcnB locus as MATERIALS AND METHODS the gene for the major Escherichia coli poly(A) polymerase Bacterial Strains and Plasmids. The E. coli strains used in (PAP I). In this report, we describe the disruption and this study were SK5691 (thyA715, pnp7) (10), SK6699 (thyA, identification of a candidate gene for a second poly(A) poly- pnp7, pcnB::kan) (8), E. coli MZ1 [his ilv rpsL galKam merase (PAP II) by an experimental strategy which was based pglA8(bio-uvrB)AH1k' (cI857 N rex)] (11). E. coli InvaF' on the assumption that the viability of E. coli depends on the [endAl recAl hsdR17 (rj mjt) supE44 A- thi-1 gyrA96 relAl presence of either PAP I or PAP II. The coding region thus F80 lacZaAM15 A (lacZYA-argF)U169)] (Invitrogen) and E. identified is the open reading frameJ310, located at about 87 coli BL21 (DE3) [F- ompT (rjm-) DE3] (12) Strains were min on the E. coli chromosome. The following lines of evidence grown in Luria-Bertani medium and antibiotics were used at support J310 as the gene for PAP II: (i) the deduced peptide the following concentrations: 50 ,ug/ml ampicillin, 50 ,ug/ml encoded byj310 has a molecular weight of 36,300, similar to kanamycin, and 10 ,ug/ml chloramphenicol. the molecular weight of 35,000 estimated by gel filtration of Plasmid vector pBAD18 (13), containing the araBAD pro- PAP II; (ii) the deduced J310 product is a relatively hydro- moter and the regulatory gene araC, was the gift of Luz-Maria phobic polypeptide with a pl of 9.4, consistent with the Guzman (Harvard Medical School). pRE1 (14), a vector de- properties of partially purified PAP II; (iii) overexpression of signed for the controlled expression of lethal genes, was the gift J310 leads to the formation of inclusion bodies whose solubi- of P. Reddy (National Institute of Standards and Technology). lization and renaturation yields poly(A) polymerase activity pRE1-pcnB (designated pRE1-1 in ref. 2), a pRE1 derivative that corresponds to a 35-kDa protein as shown by containing the entire pcnB coding region, was described earlier blotting; and (iv) expression of a J310 fusion construct with (2). pCRII version 2.2. a vector for the direct cloning of PCR hexahistidine at the N-terminus of the coding region allowed products, was from Invitrogen. pET28a, a vector for the N- purification of a poly(A) polymerase fraction whose major terminal fusion of a gene product to hexahistidine and expression component is a 36-kDa protein. E. coli PAP II has no signif- from a T7/lac promoter, was from Novagen. ANK1324 (15), a icant sequence homology either to PAP I or to the viral and vehicle for the delivery of a mini-TnlO carrying the cat gene of eukaryotic poly(A) , suggesting that the bacterial pACYC184, was obtained from Jon Beckwith (Harvard Medical poly(A) polymerases have evolved independently. An interest- School). ing feature ofthe PAP II sequence is the presence of sets of two Materials. Restriction endonucleases were from New En- paired cysteine and histidine residues that resemble the RNA gland BioLabs; Taq DNA polymerase from Perkin-Elmer/ binding motifs seen in some other proteins. Cetus; L-arabinose from Pfanstiehl Laboratories; His-bind metal chelation resin from Novagen; [3H]ATP from Moravek The recent identification of the pcnB locus (1) as the gene for Biochemicals (Brea, CA); and poly(A) from Pharmacia. Oli- the major poly(A) polymerase (ATP:polyribonucleotide ade- gonucleotides were synthesized by the phosphoramidite nylyltransferase, EC 2.7.7.19) of Escherichia coli (2) has method with a MilliGen Expedite DNA synthesizer. opened the door to the analysis of the function and mechanism Poly(A) Polymerase Assay. Poly(A) polymerase was assayed as of RNA polyadenylylation in prokaryotes on the molecular described (8). Reaction volumes of 50 ,ul were used, containing level. However, although it is clear that the poly(A) polymer- 50 mM Tris HCl (pH 8.0), 10 mM MgCl2, 2 mM MnCl2, 1 mM ase (PAP I) encoded bypcnB is essential for the 3'-adenylation EDTA, 1 mM DTT, 100 mM NaCl, 80 mM KCl, 5 ,ug/ml of the antisense RNAs that control the replication of certain rifampicin, 1 mM phosphoenol pyruvate, 3 jig ofpyruvate , plasmids (3-6), the observation that disruption of the pcnB 25 ,ug of gelatin, 0.12 mM [3H]ATP (50 ,uCi/,umol; 1 Ci = 37 locus by a mini-kan insertion (7) leads only to a modest GBq), and 10 ,tg poly(A) primer. After 10 min at 37°C, poly(A) reduction in the level of mRNA polyadenylylation (2) suggests was adsorbed onto DE-81 filter discs and radioactivity was that E. coli has more than one poly(A) polymerases. Indeed, determined in a liquid scintillation spectrometer. we were able to identify a second poly(A) polymerase (PAP II) Enzyme Blot Analysis. E. coli MZ1 transformed with pRE1, in extracts of E. coli with a pcnB deletion (8). pRE1-f310, or pRE1-pcnB were grown in medium E (16) To elucidate the function of polyadenylylation of mRNA in supplemented with 0.8% Difco nutrient broth, 0.5% glucose, E. coli, it is essential to define the relative roles of the two and 25 ,tg/ml ampicillin at 30°C to A650 of 0.55, and were poly(A) polymerases in RNA metabolism and processing. This induced for 1 -h at 39°C. Cells were disrupted in a French can be accomplished only if the genes for both are pressure cell, and aliquots of the crude extract (3 ,ul), together identified. In this paper, we describe our successful strategy for with a mixture of prestained proteins of known molecular disrupting the gene for PAP II in the absence of a functional weight (Bio-Rad), were subjected to SDS/PAGE in 12.5% chromosomal gene for PAP I, which has led to the identifi- polyacrylamide gels, followed by electroblotting onto Problot cation of the previously unassigned open reading frame f310 membrane (Applied Biosystems) in 0.1 M 3-(cyclohexylamino)- (9) as the gene encoding PAP II. propanesulfonate buffer (pH 11.0) containing 20% methanol. The membrane was screened for poly(A) polymerase activity The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in Abbreviations: PAP I and II, poly(A) polymerase I and II; IPTG, accordance with 18 U.S.C. §1734 solely to indicate this fact. isopropyl P3-D-thiogalactoside. 11580 Downloaded by guest on September 25, 2021 Biochemistry: Sarkar et al. Proc. Natl. Acad. Sci. USA 93 (1996) 11581 by incubation for 16 h at 23°C in 7 ml of a solution containing overexpression of the gene thus identified and assay for 50 mM Tris-HCl (pH 8.0), 180 mM NaCl, 10 mM MgCl2, 2 mM poly(A) polymerase activity. MnCl2, 2 mM EDTA, 1 mM DTT, 160 ,ug/ml streptolydigin, Construction ofan E. coli Strain withpcnB Under Arabinose 10 mM phosphocreatine, 100 ,tg/ml , 0.12 mM Control. The starting point for our construction was E. coli [a-35S]ATP (100 ,uCi/,tmol), and 70 ,tg poly(A) primer. The strain SK6699 (8), which lacks the gene for polynucleotide membrane was then washed at 4°C twice with 1.2 M trichlo- (pnp) and has a disrupted PAP I locus roacetic acid containing 2.5 mM ATP, four times with 0.6 M (pcnB::kan). The aim was to introduce into this host a plasmid trichloroacetic acid, and once with water, dried at room vector carrying an arabinose-inducible pcnB gene. The vector temperature, and subjected to autoradiography using a storage of choice was pBAD18 (13), which confers ampicillin- phosphor screen and analyzed on a Molecular Dynamics resistance and contains the araBAD promoter (PBAD) and Phosphorlmager SF with IMAGEQUANT software. araC, a tight regulator of PBAD which allows expression only in Isolation and Renaturation of PAP II from Inclusion Bod- the presence of arabinose. To insertpcnB adjacent to the PBAD ies. E. coli MZ1 harboring either pRE1 or pRE1-f310 were promoter, we used PCR to amplify this gene from plasmid grown in medium E (16) supplemented with 0.8% Difco pREI-1 (2), using primers chosen so as to allow amplification nutrient broth, 0.5% glucose, and 25 ,tg/ml ampicillin at 30°C of the complete coding sequence for PAP I. The PCR product to A650 of 0.55, and were induced for 1 h at 39°C. Cells from was inserted into pBAD18 to yield pBAD18-pcnB. When E. 250 ml culture were harvested and disrupted in 2 ml of buffer coli SK6699, which had only 20% of the normal poly(A) E (50 mM Tris-HCl, pH 7.9/200 mM KCl/10 mM MgCl2/1 polymerase levels of the parent strain (E. coli SK5691), was mM DTT/1 mM EDTA/5% glycerol/0.5 mM phenylmethyl- transformed with pBAD18-pcnB, poly(A) polymerase levels in- sulfonyl fluoride) in a French Pressure cell. Inclusion bodies, creased 35-fold upon induction with arabinose (data not shown). together with other particulate material, were sedimented by Transposon Mutagenesis ofE. coli SK6699/pBAD18-PAPI. centrifuging the crude cell lysate at 20,000 x g for 15 min. To disrupt the gene for PAP II in the E. coli strain SK6699/ Pellets were washed twice with buffer E by dispersing and pBAD18-pcnB, we used the minitransposon TnlO vehicle, recentrifuging the suspension. The washed pellet was extracted ANK1324 (15), and selected chloramphenicol-resistant colo- by homogenizing manually every 10 min with 0.5 ml of nies by plating phage-infected cells on agar containing ampi- denaturation buffer, 6 M urea containing 20 mM Tris HCI (pH cillin (to maintain pBAD/PAPI), chloramphenicol, and arab- 8.5), 2 mM EDTA, and 10 mM DTT for 1 h at 4°C. The inose. Ten thousand colonies were then screened for growth suspension was centrifuged and the solubilized proteins were on two kinds of plates containing Luria-Bertani agar with both renatured as described by Nagai et al. (17) by dialysis against antibiotics and either 0.02% arabinose or none. A single 100 ml of the first dialysis buffer (3 M urea/20 mM Tris HCl ampicillin- and chloramphenicol-resistant colony, T99, was pH 8.0/200 mM KCl/5 mM DTT/1 mM EDTA) for 14 h at found which grew only in the presence of arabinose but not in 4°C and then against 200 ml of the second dialysis buffer (20 its absence, the anticipated phenotype of a strain with mini- mM Tris-HCl pH 8.0/200 mM KCl/5 mM DTT/1 mM EDTA/ cam disruption of PAP II gene. The phenotype of colony T99 10% glycerol) for 8 h. was confirmed by streaking on Luria-Bertani agar plates with Purification and Renaturation of N-Terminal Hexahisti- the two antibiotics and with or without 0.02% arabinose, but dine-Tagged PAP II from Inclusion Bodies. E. coli strain all subsequent efforts to subculture this strain, either on plates BL21(DE3), transformed with pET28a or pET28a-f310, was or in liquid media, were unsuccessful. In hindsight, such loss of grown in Luria-Bertani medium plus kanamycin (30 ,ug/ml) to viability should have been expected, because arabinose, at the A540 of 0.6, then induced for 3 h by the addition of 1 mM low level (200 ,tg/ml) used in the agar, would have been isopropyl f3-D-thiogalactoside (IPTG). Harvested cells (110 mg substantially depleted around the bacterial colonies, curtailing wet weight) were resuspended in 2 ml of buffer B (20 mM the synthesis of PAP I. This could have been avoided by Tris-HCl, pH 7.9/500 mM NaCl/5 mM imidazole) and cells subculturing before the colonies had reached their maximum were disrupted in a French pressure cell. Inclusion bodies were size, as was done in the original screening. collected by centrifugation at 20,000 x g for 15 min and Amplification and Identification of the Transposon- dispersed in 0.25 ml of buffer B containing 6 M urea. After 1 h Disrupted Gene. The transposon-disrupted gene from the at 0°C to allow for solubilization, the hexahistidine-tagged nonviable arabinose-dependent T99 colony was recovered by protein was purified by batch-mode metal-ion-affinity chro- inverse PCR. Chromosomal DNA (5 gg) from cells of colony matography. The sample was absorbed onto 0.2 ml of Ni2+- T99 scraped from the agar plate was partially digested with Sepharose (Novagen) in buffer B with 6 M urea. The resin was Sau3AI, followed by religation at a relatively low DNA con- spun down and washed twice with 1 ml of buffer B with 6 M centration (1.25 ng/ml) to encourage circularization. The urea and twice with 1 ml ofbinding buffer B with 6 M urea and circular DNA was then amplified by inverse PCR, using 20 mM imidazole. Bound proteins were then eluted by 0.4 ml primers complementary to the 70-bp inverted repeats at the of buffer B with 6 M urea and 300 mM imidazole. The eluted ends of mini-TnlO (cat), which conveniently contain Sau3A1 protein was renatured by the two-step dialysis procedure of sites outside the primer sequence selected (15), as described by Nagai et al. (17) as described above. Higashitani et al. (18). The inverse PCR products were ligated Other Methods. DNA amplification by PCR and DNA to vector pCRII and used to transform E. coli InvaF'. Re- sequencing were carried out as described (2). combinant clones with inserts were identified and the nucle- otide sequence of the chromosomal regions adjacent to the RESULTS inverted repeats at the end of TnlO insert of two of the clones, Strategy for Identifying the Gene for a Second Poly(A) T4 and T8, were determined. Polymerase. Our experimental approach was predicated on Sequences of50 and 94 bp were obtained from clones T4 and the assumption that the absence of both PAP I and PAP II is T8, respectively, and used to search for homologous segments lethal to E. coli. Accordingly, we developed the following in the E. coli DNA data base. Both sequences found exact four-step strategy for identifying the PAP II gene: (i) deletion matches in a region near 87 min on the E. coli DNA data base, of the chromosomal pcnB locus and introduction of a plasmid sequenced earlier by Blattner and coworkers (9). As shown in copy of pcnB under control of the arabinose promoter; (ii) Fig. 1, the two matches were in the same open reading frame transposon mutagenesis and isolation of arabinose-dependent termed 310, which had not yet been assigned a function and transposon mutants; (iii) amplification of the transposon- whose deduced product is a 310-residue protein of Mr 36,300, disrupted locus by inverse PCR, followed by cloning, sequenc- in reasonable agreement with the estimated Mr of 35,000 for ing, and comparison to the E. coli genome data base; and (iv) the second poly(A) polymerase of E. coli (8). Downloaded by guest on September 25, 2021 11582 Biochemistry: Sarkar et aL Proc. Natl. Acad. Sci. USA 93 (1996)

TCATGCAGTTAAAAATAAATCCATCCCTAATAATTGAGATAATGGGATTTTTTCTAATGTGGATAGATATGAATTATT 78 I I

TTTCTCCTTAAGGATCATCCGTTATTTGGGTCGTTTTCTGCTAAAATCTGCGCCTTGCGCTTCTTTAGCGCACTGTCT 156 F L L K D H P L F G S F S A K I C A L R F F S A L S

ATGGCTAACCTTTTGAATAAATTCATTATGACGAGAATACTCGCTGCGATAACCCTTTTGCTGAGTATCGTATTGACC 234 M A N L L N KF I M T R I L A A I T LLL S I V L T ATTTTGGTCACTATTTTTTGTTCCGTCCCGATCATCATTGCCGGGATTGTAAAACTATTGCTACCTGTGCCAGTCATT 312 I L V T I F C S V P I I I A G I V K LLL P V P V I TGGCGAAAGGTTTCGCGTTTTTGTGATTTTATGATGTATTGCTGGTGTGAAGGTCTGGCGGTATTACTGCATCTGAAC 390 W R K V S R F C D F M M Y C W C EG L A V LL H L N

CCACACCTTCAATGGGAAGTTCACGGGCTGGAAGGGTTAAGTAAGAAGAACTGGTATCTGCTTATTTGTAACCACCGT 468 P HL 0 W E V H G L E G L S K K N W Y LL I C N H R (C?) AGTTGGGCAGATATTGTCGTACTGTGCGTGCTGTTTCGTAAGCACATTCCGATGAATAAATATTTCCTCAAGCAACAG 546 S W A D I VV L C V L F RK H I P (I?)N K Y F L K 0 0 CTGGCCTGGGTGCCATTTCTTGGCCTGGCGTGCTGGTCGCTGGATATGCCGTTTATGAAGCGTTATTCCCGCGCTTAT 624 L A W V PF L G L A C W S L D M P F M K R Y S R A Y TTATTACGTCATCCGGAGCGACGTGGTAAAGATGTTGAAACCACTCGGCGTTCTTGTGAAAAGTTTCGCCTGCATCCC 702 L L RH PE R R G K D V E T TR R S C E KFR L H P ACCACCATTGTCAATTTCGTTGAAGGCTCCCGTTTCACGCAAGAAAAACATCAGCAAACCCACTCCACTTTTCAAAAC 780 T T I V N F V E G S RF T O E K H 0 0 T H S TF O N TTGTTGCCACCAAAGGCGGCAGGCATTGCGATGGCGCTAAATGTACTGGGTAAACAATTCGATAAACTGCTTAATGTT 858 L L P P K A A G I A M A L N V L G K O F D K LL N V ACGCTGTGTTATCCGGACAATAACCGTCAGCCGTTCTTCGATATGTTAAGCGGTAAACTGACGCGGATTGTCGTACAT 936 T L C Y P D N N R 0 P FF D M L S G K L T R I VV H GTGGATTTACAGCCTATTGCCGATGAGTTACATGGCGACTACATCAATGACAAAAGCTTTAAGCGCCATTTTCAGCAA 1014 V D L Q P I A D E L H G D Y I N D K S F K R HF 0 0 TGGCTGAACTCATTGTGGCAGGAAAAAGACCGGCTTCTGACGTCTCTCATGTCATCACAGCGTCAGAATAAGTAAGTT 1092 W L N S L W 0 E K D R L LT S L M S S O R O N K TTACTCGCCTCTGAATGAGCAGAGGCGAGTGAGTATTTTAATGAACAAAACGTCCGGCACGAGACATAAATTCTTCTT 1170 FIG. 1. and deduced amino acid sequence off310, the putative gene for PAP II. The sequence shown corresponds to 8398-7229 of the E. coli chromosomal region from 87.2 to 89.2 min (ref. 9; GenBank accession no. L19201), with the complementary sequence being shown; 1 corresponds to 8398 of L19201. The deduced amino acid sequence of PAP II is shown, with amino acids corresponding to codons upstream of the most likely translation initiation site at the first methionine residue (residue 157) in italics. The nucleotide sequences clones T4 and T8 obtained after inverse PCR are underlined, with the Sau3Al sites doubly underlined. Note that the G residue originally reported (9) at position 522 of L19201 is probably in error and should be a C so as to constitute a Sau3Al site at this position. Expression of Poly(A) Polymerase Activity from the Cloned blotting-enzyme staining procedure. Equivalent amounts of 1310 Gene. To determine whether f310 indeed encodes a crude extracts from cells transformed with pRE1-J310, with cells poly(A) polymerase, we amplified the f310 gene from E. coli transformed with nonrecombinant pRE1, and with pRE1-pcnB chromosomal DNA by PCR using mismatched primers to as negative and positive controls, respectively, were subjected to generate an NdeI site near the ATG initiation codon and a SDS/PAGE and blotted on polyvinylidene difluoride mem- BamHI site after the TAA stop codon. The amplified fragment branes, followed by incubation with poly(A) primer and was ligated into the pRE1 expression vector, and the recom- [a-35S]ATP, washing with trichloroacetic acid, and autoradiog- binant plasmid, pRE1-f310, was used to transform E. coli MZ1. raphy. As shown in Fig. 2B, radioactive bands indicative of The transformed strain, as well as a control strain with poly(A) synthesis were seen with extracts from cells transformed nonrecombinant pRE1, were grown at 30°C and APL expres- with pRE1-1310 and pRE1-pcnB, but not in extracts from cells sion was induced by temperature shift to 39°C for 1 h. Extracts transformed with the negative control plasmid, pREL. The from induced and uninduced cultures from the two strains estimated molecular weights of the poly(A) polymerizing activi- were prepared by disruption with a French pressure cell and ties that resulted from overexpression of pRE1-J310 and pREl- assayed for poly(A) polymerase activity either directly or after pcnB were 35,000 and 50,000, in close agreement with the centrifugation at 100,000 x g. No differences in poly(A) molecular weights of PAP II and I, respectively. polymerase activity were seen in the extracts from cells Attempts were made to solubilize the inclusion bodies transformed with either pRE1-f310 or the control plasmid formed upon overexpression of the f310 gene in cells trans- pRE1 (data not shown). Analysis of the same extracts by formed with plasmid pRE1-f310. The material which sedi- SDS/PAGE showed an overexpressed 36-kDa protein in the mented at 20,000 x g was treated with 6 M urea containing 20 strain transformed with pRE1-f310, which was completely mM Tris HCl (pH 8.5), 10 mM DTT, and 2 mM EDTA. The sedimented at 100,000 x g, presumably as inclusion bodies (see solubilized protein was then allowed to refold by slowly Fig. 3A). To determine whether poly(A) polymerase activity removing the denaturant by a two-stage dialysis procedure was associated with the overexpressed 36-kDa protein, we used a (17). Poly(A) polymerase activity was then measured in the Downloaded by guest on September 25, 2021 Biochemistry: Sarkar et aL Proc. Natl. Acad. Sci. USA 93 (1996) 11583 PAGE showed that the solubilized material from cells with A B pRE1-J310 was highly heterogeneous (data not shown). Purification and Renaturation of Overexpressed PAP II as c:>Q a Hexahistidine Fusion Protein. To ascertain that the regen- erated activity is due to the overexpression of a single polypep- tide encoded the we recloned the Q by f310 gene, f310 gene Lu uwW sequence into expression vector pET28a so as to fuse six Z. 'mC consecutive histidine residues to the N terminus of the ex- kDa CL CL kDa _L pressed protein and allow its purification by chelation to a Ni2+ 101 - column (19). The f310 sequence was excised from plasmid - 96 - 83 pRE1-f310 with NdeI and BamHI and ligated to vector pET- - - 55 51 28a cut with the same restriction enzymes, and the recombi- nant plasmid, pET-28a-f310, was used to transform E. coli -43 36- BL21 (DE3). A parallel control experiment was done with E. r- 36 coli BL21 (DE3) transformed with nonrecombinant pET28a. - 24 29 - Expression was induced by adding 1 mM IPTG to exponen- tially growing cultures of the transformed strain. Examination of the extracts of the induced cells by SDS/PAGE revealed the - 18 21- presence of overexpressed 36-kDa protein in the insoluble - 12 fraction but not in the supernatant, consistent with the for- mation of inclusion bodies (Fig. 4A). Exploiting the His-Tag sequence at the N terminus of the overexpressed protein, we purified the overexpressed protein from the insoluble fraction FIG. 2. SDS/PAGE and enzyme blot analysis oIfproteins produced by solubilization in 6 M urea, followed by affinity chromatog- by E. coli strains transformed with pRE1 carryingrthe f310 gene. The raphy as described in Materials and Methods. As shown in Fig. transformed cells were grown and disrupted and thie cell extracts were 4B, the polypeptides which eluted from the Ni2+ column with processed as described. SDS/PAGE was carried out in 12.5% poly- 1 M imidazole buffer consisted of single major 36-kDa band, acrylamide gels together with a mixture of proteiris of known molec- together with two minor bands of higher molecular weight. In ular weight, followed by staining with Coomassie blue. The overex- view of the fact that PAP II contains nine cysteine residues and pressed 36-kDa protein is indicated by arrows. (A) Crude cell extracts purification had to be carried out in thiol-free buffers, it is (10 ,ug of protein) from E. coli MZ1 transfornned with pRE1 or possible that the latter may represent disulfide-linked multim- pRE1-f310 and induced for 1 h at 39°C. (B) Crude cell extracts (3 from E. coli MZ1 transformed with pRE1, pRE1-] pc) ers of the fusion protein or co-aggregates with other proteins and induced for 1 h at 39°C were subjected to SDS,/PAGE as inA and (20). A 36-kDa component could not be recognized in the then blotted onto a Problot membrane and asssayed for poly(A) corresponding fractions from cells transformed with nonre- polymerase activity as described. pRE1-pcnB, whiich carries the gene combinant pET28a. The purified polypeptide fractions from for PAP I (2), was included as a positive control. cells transformed with pET28a-J310 as well as from control cells were renatured by a two-stage dialysis procedure (17) and renatured extracts, with similar extracts from E. coli MZ1 poly(A)-polymerase activity was measured. Substantial containing pRE1 vector alone as a control. R.enatured extracts poly(A) polymerase activity was found in the purified protein from the insoluble fraction of E. coli MZ1/pRE1-f310 had an fraction from cells transformed with the vector carrying the activity of 2 units/mg protein, whereas an;analogous prepa- f310 gene but not in the corresponding fraction from control ration from the host strain transformed with 1the control vector cells (Fig. 3B), suggesting that the f310 gene encodes a had one-fifteenth as much activity (Fig. 3A)*. However, SDS/ polypeptide with poly(A) polymerase activity.

120 25 E 0E o 100 0 0 E &fi 80 U ._s 0 =2 0 0c: 6 0 0 6 SCL

84 0 Eco a0 8> < 20 0.0 0

0 10 20 30 40 50 60 0.0 0.2 0.4 0.6 0.8 Amount of protein (gg) Amount of protein (jig) FIG. 3. Assay of poly(A) polymerase activity in extracts of E. coli strains transformed with expression vectors carrying the f310 gene. (A) The insoluble fraction from induced cells of E. coli MZ1 transformed with pRE1 (0) or pRE1-f310 (0) was solubilized with 6 M urea, renatured, and assayed for poly(A) polymerase activity as described. (B) The insoluble fraction from IPTG-induced cells of E. coli BL21 (DE3) transformed with pET28a (0) or pET28a-f310 (-) was solubilized, fractionated on Ni2+-Sepharose, and assayed for poly(A) polymerase activity as described. Downloaded by guest on September 25, 2021 11584 Biochemistry: Sarkar et al. Proc. Natl. Acad. Sci. USA 93 (1996) A B similar to the molecular weight of 35,000 estimated by gel- pET28 filtration of PAP 11 (8); (ii) the deduced f310 product is a pET28a pET28a-f31 0 pET28a -f31 0 relatively hydrophobic peptide with a pl of 9.4, consistent with the properties of partially purified PAP II, which binds strongly 0 0 en X co to to to 3 co hydrophobic matrices such as phenyl agarose, eluting only _U . ~: o z 3: I with difficulty and in low yield (8), and which adsorbs to LOO-a i 0 ^ O9 sulfopropyl Sepharose but not to DEAE Sepharose, at neutral kDa pH (M. Kalapos and N.S., unpublished results); (iii) overex- -96 pression off310 in appropriate expression vectors led to the formation of inclusion bodies whose solubilization and rena- -55 turation yielded substantial poly(A) polymerase activity, (iv) SDS/PAGE of the proteins from cells transformed with - 43 - 36 pRE1-f310 followed by enzyme blotting revealed a single ^-24 overexpressed poly(A) polymerase activity with a molecular weight or 35,000; and (v) expression of af310 fusion construct with hexahistidine at the N. terminus of the coding region allowed purification of a poly(A) polymerase fraction whose -18 major component (85%) was a 36-kDa protein. The specific -12 activity of the purified poly(A) polymerase fusion protein was 40 units per mg of protein, about 25% of that of the 1000-fold FIG. 4. SDS/PAGE analysis of proteins produced by E. coli strains purified preparation of PAP II obtained earlier (8). The lower transformed with an expression vectors carrying a f310 gene- specific activity of the fusion protein could have been due to hexahistidine fusion. The transformed cells were grown and disrupted some irreversible inactivation during denaturation of the in- and the cell extracts were processed as described. SDS/PAGE was clusion bodies and metal ion chromatography, which had to be carried out in 12.5% polyacrylamide gels together with a mixture of done in the absence of thiols, considering that PAP II is a proteins of known molecular weight, followed by staining with Coo- massie blue. The overexpressed 36-kDa protein is indicated by arrows. relatively unstable enzyme (8), and perhaps also to the pres- (A) Extracts of IPTG-induced cells ofE. coli BL21 (DE3) transformed ence of the N-terminal hexahistidine moiety, whose effect on with pET28a or pET28a-f310, before and after fractionation into a enzyme activity cannot be predicted. soluble and inclusion body (pellet) fractions by centrifugation at The deduced 310-residue polypeptide product off310 has a 20,000 x g for 15 min. (B) The inclusion body (pellet) fractions from relatively high percentage of hydrophobic amino acids, many of A were solubilized and fractionated on Ni2+-Sepharose, and equal which are concentrated in the 50 N-terminal residues as shown by proportions of the unbound (Ni wash) and the bound and eluted (Ni the hydrophilicity plot (Fig. 5). The high degree ofhydrophobicity eluate) fractions were examined by SDS/PAGE. and the clustering ofthe hydrophobic residues may be responsible for the tendency of the overexpressed product of thefl0 gene to DISCUSSION form inclusion bodies. Another notable feature is the relatively In this report, we describe the disruption and identification of large excess of basic over acidic amino acids; as a result, the a candidate gene for a second poly(A) polymerase (PAP II) by protein has a predicted pl of 9.4. Comparison of the deduced an experimental strategy that was based on the assumption amino acid sequence of the J310 gene product with the protein that the viability of E. coli depends on the presence of either sequence data banks revealed no significant homology with any PAP I or PAP II. The gene thus identified was the open known protein. On the other hand, search for possible motifs with reading framef310, which is located at about 87 min on the E. functional significance uncovered two interesting features. One of coli chromosome adjacent topoL4 but oriented in the opposite these are two sets of paired cysteine and histidine residues: direction (9). The following lines of evidence support the idea residues 67-80, CWCEGLAVLLHLNPH; and residues 101-119, thatJ310 is indeed the gene for PAP II: (i) the deduced peptide CNHRSWADIVVLCVLFRKH. These clusters of cysteine and encoded by f310 has a molecular weight of 36,300, which is histidine residues are reminiscent of the Zn2+-binding domains

I I I I I , i- Amino acid residues 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 3.29 A ji,it L * , 0 _ r n w w w S S wT~ Hydrophilicity Plot _ w | r 1 r~~~w1 I

+ i i1 i iIH I H 11 Positive Charges

i i i i i i i i i ill l 1 l IIR Negative Charges Zn finger? Zn finger? IE.ZI LZ_.J

a _ 1. M. I I Alpha Regions B _I I _I -.m m_-*m m in11 m Im U i U - I: i Beta Regions i i T Eli i .i i. i 1 11I n -i I11 Turn Regions

FIG. 5. Structure predictions for PAP II, the deduced f310 gene product. Hydrophilicity was estimated by the method of Kyte and Doolittle (21) and the secondary structure predictions were based on the method of Garnier et al. (22), utilizing the LASERGENE software package of DNAstar (Madison, WI). Also shown are the disposition of basic and acidic amino acid residues and the location of the putative zinc fingers. Amino acids are numbered starting at the methionine at nucleotide 157 (Fig. 1). Downloaded by guest on September 25, 2021 Biochemistry: Sarkar et al. Proc. Natl. Acad. Sci. USA 93 (1996) 11585 often seen in nucleic acid-binding proteins and in hormone also considerable redundance in the mRNA polyadenylylating receptor proteins and referred to as zinc-fingers (23). The fact function of the two E. coli poly(A) polymerase, since disrup- that the amino acids surrounding the cysteine/histidine pairs are tion of thepcnB gene encoding PAP I has only modest effects not strictly in accord with the consensus sequence characteristic on the extent of mRNA polyadenylylation (8). Indeed, redun- of the zinc fingers of DNA binding proteins (24), especially with dancy with respect to enzymes acting on RNA seems almost to respect to spacing of the first Zn2+-binding pair, may be related be the rule in E. coli. For eXample, 3'-exonucleolytic degra- to different structural requirements for the interaction ofpoly(A) dation of mRNA can be effected both by polynucleotide polymerase with a single strand of RNA. In this connection, it is phosphorylase and RNase II, so that mutational inactivation of interesting that the zinc-binding domain of the glutamyl tRNA either enzyme can be tolerated, but not the loss of both (10). synthetases of E. coli and several other bacterial species, which Similarly, there is a high degree of functional overlap among has recently been shown to be involved in tRNA binding (25), the 3' involved in tRNA maturation, which consists of the sequence CYCX24CRHI (26). Another example of include RNase D, RNase BN, RNase T, and RNase PH (33). a zinc finger with only a single amino acid residue between the The recent discovery of rnhB, a gene encoding a second upstream cysteine pair is zinc finger 3 ofyeast transcription factor H (34) is another example of such functional SW15 (27). The other feature of the deduced f310 translation duplication. Most likely, such redundancy serves in part as a product that resembles a known structural motif is the hydro- fail-safe mechanism to guard. against the loss of a vital func- phobic N-terminal segment. The 12 N-terminal residue, which tion. Ironically, this functional and genetic redundancy has include 2 basic amino acids, are followed by 9 nonpolar residues greatly hampered the-genetic analysis of mRNA metabolism in (67% alanine plus leucine) with a-helical tendency (Fig. 5), E. coli and is primarily responsible for our ignorance of this similar to the "N" and "H" domains of the N-terminal signal vital metabolic area. The identification of the gene for the peptides of the bacterial secretory protein precursors (28). How- second E. coli poly(A) polymerase opens the way for the ever, whereas the signal peptide "H" domain is generally followed detailed investigation of the metabolic role of mRNA polya- by a "C" domain, which contains the recognition sequence for denylylation by studying the consequences of disruption of signal peptidases, and then by a stretch of hydrophilic and acidic either or both of the poly(A) polymerase genes. amino acids, theJ310 product continues with almost 40 additional hydrophobic and basic amino acid residues. It is therefore unlikely We thank Drs. Jon Beckwith and Henry Paulus for stimulating that this extensive hydrophobic and cationic N-terminal domain discussions and critical reading of the manuscript. This work was is a signal sequence for protein secretion and its function in the supported by National Institutes of Health Grant R01-GM26517. context of poly(A) polymerase remains to be elucidated. 1. Lopilato, J., Bortner, S. & Beckwith, J. (1986) Mol. Gen. Genet. 205, A common feature of the genes encoding PAP I and PAP 285-290. 2. Cao, G.-J. & Sarkar, N. (1992) Proc. Nati. Acad. Sci. USA 89, 10380-10384. II are relatively weak transcription or translation initiation 3. He, L., Soderbom, F., Wagner, E. G. H., Binnie, U., Binns, N. & Masters, elements, which may serve to keep the levels of expression low M. (1993) Mol. Microbiol. 9, 1131-1142. to avoid possible toxic effects of the kind observed when PAP 4. Xu, F., Lin-Chao, S. & Cohen, S. N. (1993) Proc. Natl. Acad. Sci. USA 90, 6756-6760. I is overexpressed (2). Examination of the f310 sequence 5. Xu, F. & Cohen, S. N. (1995) Nature (London) 374, 180-183. suggest the most likely translation start site for PAP II to be 6. Cohen, S. (1995) Cell 80, 829-832. the AUG codon at position 157 (Fig. 2), which, although 7. Liu, J. & Parkinson, J. S. (1991) J. Bacteriol. 173, 4941-4951. 8. Kalapos, M. P., Cao, G.-J., Kushner, S. R. & Sarkar, N. (1994) Biochem. followed by the favorable sequence GCUA (29), is not pre- Biophys. Res. Commun. 198, 459-465. ceded by a discernible ribosome . (Possible alter- 9. Plunkett, G., III, Burland, V., Daniels, D. L. & Blattner, F. R. (1993) nate translation initiation codons, such as UUG at position 134 Nucleic Acids Res. 21, 3391-3398. 10. Arriano, C. M., Yancey, S. D. & Kushner, S. R. (1988) J. Bacteriol. 170, or AUG at position 184 are also not associated with recog- 4625-4633. nizable ribosome binding sites.) In comparison, PAP I syn- 11. Zuber, M., Patterson, T. A. & Court, D. L. (1987) Proc. Natl. Acad. Sci. thesis starts with the unfavorable UUG initiation codon and is USA 84, 4514-4518. 12. Studier, F. W., Rosenberg, A. H., Dunn, J. J. & Dubendorff, J. W. (1990) preceded by a very weak ribosome binding site (2). As far as Methods Enzymol. 185, 60-89. transcription initiation of1310 is concerned, no sequence that 13. Guzman, L. M., Belin, D., Carson, M. J. & Beckwith, J. (1995) J. Bacteriol. fits the canonical -10 and -35 elements for the major E. coli 177, 4121-4130. 14. Reddy, P., Peterkofsky, A. & McKenney, K. (1989) Nucleic Acids Res. 17, RNA polymerase can be discerned upstream of the putative 10473-10488. coding region, even though transcription must start betweenf310 15. Kleckner, N., Bender, J. & Gottesman, S. (1991) Methods Enzymol. 204, and the beginning of the divergently transcribed poLA gene. It is 139-180. interesting that the E. coli dnaG gene, which encodes RNA 16. Vogel, H. J. & Bonner, D. M. (1956) J. Biol. Chem. 218, 97-106. 17. Nagai, K., Nakasuko, Y., Nasmyth, K. & Rhodes, D. (1988) Nature , a protein produced only in very small amounts, also lacks (London) 332, 284-286. discernible promoter and ribosome binding sites (30). 18. Higashitani, A., Higashitani, N., Yasuda, S. & Horiuchi, K. (1994) Nucleic There is no sequence homology between the two E. coli Acids Res. 22, 2426-2427. 19. Gentz, R., Chen, C.-H. & Rosen, C. A. (1989) Proc. Natl. Acad. Sci. USA poly(A) polymerases, PAP I and PAP II, nor between the E. 86, 821-824. coli enzymes and those from eukaryotes and viruses, suggest- 20. Hengen, P. N. (1995) Trends Biochem. Sci. 20, 285-286. ing independent origins and convergent evolution in terms of 21. Kyte, J. & Doolittle, R. F. (1982) J. Mol. Biol. 157, 105-132. 22. Garnier, J., Osguthorpe, D. J. & Robson, B. (1978) J. Mol. Biol. 120, function. Indeed, PAP I of E. coli, besides being capable of 97-120. polyadenylylating mRNA (31, 32), also functions in the repli- 23. Coleman, J. E. (1992) Annu. Rev. Biochem. 61, 897-946. cation control of plasmids of the ColEl type by transferring 24. Klug, A. & Rhodes, D. (1987) Trends Biochem. Sci. 12, 464-469. 25. Gauthier, J. & Lapointe, J. (1995) Protein Eng. 8 (Suppl.), 19. adenylate residues to the 3' ends of RNA I, the antisense 26. Liu, J., Gagnen, Y., Gauthier, J., Furenlid, L., L'Heureux, P. J., Auger, M., repressor of plasmid DNA replication, thereby accelerating its Nureki, O., Yokoyama, S. & Lapointe, J. (1995) J. Biol. Chem. 270, degradation by polynucleotide phosphorylase (4, 5). The fact 15162-15169. 27. Stillman, D. J., Bankier, A. T., Seddon, A., Groenhout, E. A. & Nasmyth, that the copy number of ColEl plasmids is greatly reduced in K. A. (1988) EMBO J. 7, 485-494. pcnB mutants (1) suggests that this regulatory function de- 28. Pugsley, A. P. (1993) Microbiol. Rev. 57, 50-108. pends uniquely on PAP I and that PAP II cannot take its place. 29. Kozak, M. (1983) Microbiol. Rev. 47, 1-45. II 30. Smiley, B. L., Lupski, J. R., Svec, P. S., McMacken, R. & Godson, G. N. The question whether PAP also has unique functions which (1982) Proc. Natl. Acad. Sci. USA 79, 4550-4554. cannot be fulfilled by PAP I cannot be answered until strains 31. O'Hara, E. G., Chekanova, J. A., Ingle, C. A., Kushner, Z. R. & Peters, E. are available in which the gene for PAP II has been specifically (1995) Proc. Natl. Acad. Sci. USA 92, 1807-1811. disrupted; the N-terminal domain of 32. Hajnsdorf, E., Braun, F., Haugel-Nielsen, J. & R6gnier, P. (1995) Proc. however, hydrophobic Natl. Acad. Sci. USA 92, 3973-3977. PAP II suggest the possibility of a membrane-associated 33. Itaya, M. (1990) Proc. Natl. Acad. Sci. USA 87, 8587-8591. function. On the other hand, there is no question that there is 34. Reuven, N. B. & Deutscher, M. P. (1993) FASEB J. 7, 143-148. Downloaded by guest on September 25, 2021