JOURNAL OF BACTERIOLOGY, Apr. 1987, p. 1663-16690021-9193/87/041663-07$02.00/0Copyright © 1987, American Society for Microbiology

Vol. 169, No. 4

Use of TnphoA to Detect Genes for Exported Proteins inEscherichia coli: Identification of the Plasmid-Encoded Gene for a

Periplasmic Acid PhosphatasePAUL L. BOQUET,l* COLIN MANOIL,2 AND JON BECKWITH2

Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, Massachusetts 021152; andService de Biochimie, Departement de Biologie, Centre d'Etudes Nucleaires de Saclay, 91191 Gif-sur- Yvette, France'

Received 23 June 1986/Accepted 8 December 1986

The structural gene (appA) for the periplasmic acid phosphatase (optimum pH 2.5) of Escherichia coli wascloned into a plasmid by using a combination of in vivo and in vitro techniques. The position and orientationof the appA gene within the cloned DNA fragment were identified by using fusions to the alkaline phosphatasegene (phoA) generated by Tn5 ISSOL::phoA (TnphoA) insertions. For TnphoA-generated hybrid proteins tohave high enzymatic activity, it appears that the phoA gene must be fused to a target gene coding for a signalwhich promotes protein export. The approach used to identify the appA gene thus appears to provide a simplegeneral means of selectively identifying genes encoding membrane and secreted proteins.

Four phosphatases have been identified in the periplasm ofEscherichia coli. These enzymes can be distinguished fromeach other by their pH optima for activity and their substratespecificities (8, 10, 19, 22, 23, 25). The periplasmic acidphosphoanhydride-phosphohydrolase (EC 3.6.1.11) referredto hereafter as acid phosphatase, is a monomeric protein ofMr 45,000 which is soluble in acid solutions (such as 1.5 Mformic acid) and optimally active at pH 2.5 against inorganicpolyphosphates, guanosine polyphosphates, and p-nitrophe-nyl phosphate (PNPP) (3, 5, 6).The expression of this acid phosphatase is subject to

complex regulation; it is influenced by the phase of growth,the presence or the absence of oxygen, and the concentra-tion of Pi in the medium. It is also inhibited by the presencein the cell of cyclic AMP and its receptor protein (5). Thestructural gene for acid phosphatase, appA, is located at 22min on the E. coli linkage map (4).To study the regulation of acid phosphatase expression,

we have cloned the appA gene. We identified the positionand orientation of this gene on a recombinant plasmid byusing fusions to alkaline phosphatase. This method shouldbe generally useful in identifying genes for proteins exportedfrom the cytoplasm. In addition, it should be possible tostudy the regulation ofappA expression by using the alkalinephosphatase activity of appA-phoA hybrid proteins.

MATERIALS AND METHODSStraing and media. The genotypes of all strains used and

constructed and the characteristics of recombinant plasmidsand of phages are listed in Table 1. The rich medium usedwas TYE (17). The presence of transposon TnlO on thechromosome was selected with 25 pLg of tetracycline per ml.The chromogenic substrate 5-bromo-4-chloro-3-indolylphosphate was added at 40 ,g/ml for the detection of alkalinephosphatase activity on plates. Ampicillin was added at 200,pg/ml to maintain plasmid pBR322 or its derivates in strains,and chloramphenicol (25 ,ug/ml) was used to select or main-tain the "phasmid" Mu dII4042 or the recombinant plasmidsderived from it. The minimal DM medium was that of Davisand Mingioli (7). The M63 medium was that of Miller (17).

* Corresponding author.

Minimal media were supplemented with thiamine and re-quired amino acids or bases at 50 ,ug/ml. The Mu mediumwas obtained by adding MgSO4 (2.5 mM) and CaCl2 (1 mM)to TYE medium. Low-phosphate, complete TE mediumcontaining 0.5% bactopeptone was described previously (4,5).

Cloning with Mu dII4042. The in vivo cloning vehicle MudII4042 and the general cloning procedure with this plasmidhave been described by Groisman et al. (9). All steps wereperformed in strains derived from the E. coli K-12 strainK10. A stock of Mu dII4042 was obtained by thermoinduc-tion of strain SBS1073 (appA+) previously transformed withplasmid pEG109 prepared from strain Xph43 (9). This phagestock was used to transduce SBS1075 (appA Mu cts), tochloramphenicol resistance (Cm%). Cmr colonies (about 600per plate) were replica plated, and the replicas were grown at30°C and assayed for acid phosphatase with PNPP in 1.5 Mformic acid (24-26). Positive clones were purified from themaster plates and grown in liquid cultures to control the pHoptimum for PNPP hydrolysis and to measure acid phospha-tase-specific activities.

Bacteriophages and transductiong. Transductions withphage Plvir were made as described by Miller (17). Stocks ofphage Mu cts were obtained by thermoinduction of midex-ponential-phase, liquid cultures of lysogenic strains at 42°Cfor 30 min followed by further incubation at 37°C until lysis.Mixed stocks of Mu cts and Mu dII4042 were obtainedsimilarly, but the heat shock was made at 44°C, resulting inclear lysates with high transducing potency. Transductionswith Mu cts and Mu dII4042 were achieved as described byGroisman et al. (9).

General techniques for recombinant DNA analysis. PlastnidDNA was prepared by the method of Maniatis et al. (15).Restriction mapping was done by the method of Schleif andWensink (21). Transformation of competent cells was by themethod of Mandel and Higa (14). Analysis of proteinsencoded by recombinant plasmids was made from maxicells,which were prepared by the method of Sancar et al. (20) andgrown in M63 glucose minimal medium plus all amino acidsexcept methionine. [35S]methionine (74 kBq/ml) was added,and cultures were harvested after 15 min of incubation.

1663

on February 12, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

1664 BOQUET ET AL.

TABLE 1. Characteristics of strains, phages, and plasmids

Strain, phage, Genotype or description Sourceor plasmid

E. coliXph3 F- AlacUI69 trp A(brnQ-phoR)24 M. CasadabanM8820 F- araDl39 A(ara-leu)7697 A(proA-lac)XIII rpsL M. CasadabanK10 HfrC relA pitlO tonA22 T2r B. BachmanSBS1073 Same as K10, but rpsL and Mu cts This studySBS1075 Same as SBS1073, but appAl zcc: :TnJO This studyCC118 F- A(ara-leu)7697 araD139 AlacX74 galE galK AphoA20 thi rpsE C. Manoil

rpoB argE(Am) recAl appRiMRi17 F- araD139 AlacU169 rpsL thi mot relA Arbs recAS6 J. LopilatoCC102 F42 lacI3 zzf-2::TnphoAICC118 C. Manoil

BacteriophagesP1 vir Phage P1, virulent Pasteur InstitutMu cts Mu with temperature-sensitive repressor cts62 M. CasadabanMu dII4042 Mucts62 A+B ±Cmr repFISA lac(ZYA)913 M. Casadaban (9)

PlasmidspEG109 Mu dII4042::phoA proC M. Casadaban (9)pPB1101 to pPB1114 Mu dII4042: appA This study (Fig. 1)pBR322 Tcr Apr J. BeckwithpPB1131 and pPB1318 Tcs Apr: :appA(BgllI-BgII insert into BamHI site of pBR322) This study (Fig. 2)pPB1132 and pPB1301 Tcs Apr: :appA' (BamHI-BgIIl insert into BamHI site of pBR322) This study (Fig. 2)pPB1135 Tcs Apr: :appAA (Sall deletion of pPB1131) This study (Fig. 2)pPB1245 TcsApr: :appAA (ClaI deletion of pPB1131)pPB1136 Tcs Apr ::appAA (Sall deletion of pPB1132)pPB1137 Tcs Apr: :appAA (ClaI deletion of pPB1132)pPB1120 to pPB1128 Tcs Apr Kmr appA: :TnphoA This study (Fig. 4)pPB1130 Tcs Apr Kmr appA +pPB1140 to pPB1149 and Same as pPB1130 with various TnphoA locations outside appA This study (Fig. 4)pPB1160 to pPB1173

Antibody precipitation. Cells labeled with [35S]methioninewere boiled for 2 min in the presence of sodium dodecylsulfate and immunoprecipitated with polyclonal antibodiesdirected against acid phosphatase or against alkaline phos-phatase as described by Ito et al. (11).Sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

Gels containing 10 or 11% polyacrylamide were made asdescribed by Laemmli (13). For detection of radioactivity,gels were soaked in 1 M sodium salicylate, dried, and used toexpose film for 1 to 2 days at -60°C.

Isolation of fusions of alkaline phosphatase to plasmid-encoded proteins. The construction and the structure of thefusion transposon Tn5 ISSOL::phoA (TnphoA) have beendescribed in detail by Manoil and Beckwith (16). Transposi-tions of TnphoA into a given multicopy recombinant plasmidpresent in strain CC118 (F- phoA recA lacU169) wasachieved by introducing into this strain the episome F42lacI3 (zzf-2::TnphoA) from donor CC102 and selecting alka-line phosphatase-positive colonies resistant to high levels ofkanamycin on 5-bromo-4-chloro-3-indolyl phosphate plates,exactly as described previously (16). The selected cloneswere further tested for acid phosphatase activity after replicaplating.

Detection and measurement of enzyme activities. The de-tection of pH 2.5 acid phosphatase activity on replica plateswas as previously described (4, 5), except that the time ofincubation with PNPP in 1 M formic acid was reduced to 3 to5 min for the detection of clones with multiple copies of theappA gene. The determination of the enzymatic specificactivity in cells growing in liquid cultures was as describedpreviously (3-5). Alkaline phosphatase specific activity inliquid cultures was measured by incubation of toluenizedwashed cells in 150 mM Tris hydrochloride (pH 8.8) contain-

ing PNPP (25 mM). Other steps were as for acid phosphataseactivity determination. One unit of acid or alkaline phospha-tase is defined here as the amount of enzyme catalyzing thehydrolysis of 1 nmol of PNPP per min at 37°C.

RESULTS

Molecular cloning of the gene encoding acid phosphatase.The gene for acid phosphatase (appA) was originally clonedin vivo by using Mu dII4042, a mini-Mu derivative contain-ing a plasmid origin of replication (9). When Mu dII4042grows in cells, it packages chromosomal DNA lying adjacentto Mu DNA sequences, and phage particles can generateplasmids carrying such cloned chromosomal sequences afterinfection. To clone appA, Mu dII4042 was propagated on anappA+ strain (a derivative of strain K10) as described inMaterial and Methods. The resulting phage stock was usedto infect an isogenic appA- strain (SBS1075), and colonies ofcells that acquired the chloramphenicol resistance determi-nant of Mu dII4042 were screened for acid phosphataseactivity to identify those that might have acquired plasmidscarrying the appA gene.

Five independently isolated plasmids that led to higheracid phosphatase activity were subjected to restriction anal-ysis (Fig. 1). Each plasmid carried Mu dII4042 sequencesbracketing a variable length (8 to 12.5 kilobases [kb]) ofcloned chromosomal DNA. The cloned DNA fragmentspossessed a region in common (about 5.5 kb long) that ispresumed to be responsible for acid phosphatase production.Two DNA fragments carrying parts of this common regionwere subcloned from plasmid pPB1101 into pBR322 (2) forfurther analysis. Both the 5.7-kb BglII-BglIl fragment andthe 4.6-kb BamHI-BglII fragment (Fig. 1) cloned in the

J . BACTERIOL .

on February 12, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

DETECTION OF GENES FOR EXPORTED PROTEINS IN E. COLI 1665

cF| s SiS H P X PHS E H,

I I M u 1 11 CrR I11E BglT s"i

E B; Bgg; v SnBggE H

E sl I5T l

pPP1107 _ _H BT 5S

1~117 TE PESi

11I I I I I lII II II-4

5.7 kb7 4.6 kb

FIG. 1. Restriction maps ofMu dII4042 recombinant plasmids expressing high acid phosphatase activity. Mu dII4042 sequences (thick andwavy lines) bracket cloned DNA sequences. The maps are aligned to show the apparent overlap in the cloned sequences. Fragments that werefurther subcloned are indicated at the bottom. Restriction endonucleases: B, BamHI; Bg, BglIl; E, EcoRI; H, HindIll; P, PstI; S, Sall; Sm,SmaI; X, XhoI.

BamHI site of pBR322 confered high acid phosphataseactivity to cells harboring them, showing that these frag-ments carry the determinants for acid phosphatase produc-tion. However, the activities in strains harboring such re-

combinant plasmids varied according to the orientation ofthe subcloned fragments relative to the vector (Fig. 2).Attempts to subclone further the region responsible forenzyme production indicated that the SalI, ClaI, and NruIsites of the cloned fragments (Fig. 2) were positioned insequences required for the expression of acid phosphatase.

Identification of the product encoded by the cloned gene.

The high acid phosphatase activity in cells carrying plasmidpPB1132 was found to be due to overproduction of theenzyme from its cloned structural gene appA. To determinethe level of the protein responsible for acid phosphataseactivity in different strains, we made use of the unusualsolubility of acid phosphatase in 1 M formic acid (5). Formicacid extracts of cells carrying plasmid pPB1132 showed a

major protein band migrating on polyacrylamide gel electro-phoresis at the position expected for acid phosphatase (45kilodaltons). This band was not present in cells lacking theplasmid or containing the parental plasmid pBR322 (Fig. 3).

2 ..,a ow0i 2.U U

0. 0IL cL (A wu

I1 11 1,'1 1 *

Acid Phtasespec. activity(u./mg prot.)

pPB 1131 300

C1C1

11

Al 11)~)II

-; a 5

IfZ S 0.>z X

--

..0 0

0.0.v) LUIIIIIIII Ilk--Mmi- m III

U

u

Ni

3.YI I IIII 1I111

2 E 3, - e

0.U40IfloZU

iLPLr

NIr ri I II I1'NLL66"ILNWW

-; 3 a

XL G- t z -A ff

I I llIl

lIIwWI

X0.LIA ui

11.I.PPB 131810.06 kb

FIG. 2. Restriction maps ofDNA derivatives of plasmid pBR322. DNA from pBR322 was cleaved at the BamHI site and joined to eitherthe BglII-BglII or the BamHI-BglII fragment indicated in Fig. 1, yielding, pPB1131 or pPB1318 in the former case and pPB1132 or pPB1301in the latter case. The thin line represents pBR322 sequences, and the thick line represents insert sequences. Deletions referred to in the textare indicated as gaps under each plasmid. Acid phosphatase specific activities were measured for plasmids in strain CC118 (appA+). Datawere corrected by subtracting the activity due to expression from the chromosome measured in strain CC118 containing plasmid pBR322 (15U/mg of protein).

p PB 1 1 os l, a;

pP8 1110

p PB 1102

1TE BPESm BgBgCt B

IIII

pPB1J10

1kb lac

U-n0..

E

cnz SO<t cL

)) r

m-==

8011 1j III I

'E o-m W(fla ll

_~to%Z .(

Z

tlII .I.lid.- I

- m_ Er* (A umc

0o r-cU-.-Ul

-LL

)kb pPB 1301 -300

Ian I E" I I iIIII

-3000

-3000

C1

ip PB1132I 11I II

Ii

1- 6 11 II'll . 4 0% .

-V.

ac

T

ll. I 11 IJ

a_I

VOL. 169, 1987

E 6 s"M.L-

II11I

I

on February 12, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

1666 BOQUET ET AL.

a b c' pS

95o,

67T

95>

67T

47,,40o

47o

40o

29.5k

29.5k

18.4> *

13.7m *

18.4kw

1 2 3 4 5 6 1 2 3 4 1 2 3 4

FIG. 3. Identification of the product of the gene cloned on plasmid pPB1132. Proteins present in crude extracts for the 1 M formicacid-soluble fraction of cells carrying pBR322 or pPB1132, separated by sodium dodecyl sulfate-polyacrylamide electrophoresis (10%polyacrylamide) and stained with Coomassie blue are shown in panel a. Lanes 1, molecular weight standards; 2, 3, and 4, 1 M formic acidextracts corresponding, respectively to pPB1132, no plasmid, and pBR322; 5 and 6, crude extracts corresponding, respectively to pPB1132and pBR322. (b) Autoradiogram of a 10% polyacrylamide gel electropherogram of proteins from cultures of CC118(pPB1132) andCC118(pBR322) precipitated by antibodies directed against acid phosphatase. Samples correspond to an equal amount of cells. Lanes: 1,pPB1132; 2, pPB1132 treated with staphylococcal protein A envelopes without antibody; 3, CC118 without plasmid; 4, CC118(pBR322). (c)Autoradiogram of a 12% polyacrylamide gel electropherogram of proteins labeled in maxicells lysates of Mril7(pBR322) (lane 1) orMril7(pPB1132) (lane 2) and their immunoprecipitates with antibodies against acid phosphatase (lanes 3 and 4, respectively).

The level of acid phosphatase activity in the extracts corre-sponded to the amount of this protein detected. This was186,000 U per mg of protein in extracts from cells carryingpBR1132, whereas cells carrying pBR322 or without plasmid

rec -0aU-

*C

-E>x

WE- -a

-

3z E24 go-6 CfL 2.--%p

a 0w0 z

i - - 303oo]0 --H <I

showed a low specific activity (5,800 and 6,100 U/mg proteinrespectively). This background activity is presumably due tothe chromosomal appA gene. Furthermore, a heavy band ofprotein with the electrophoretic mobility of acid phosphatase

- -- - i~~-Z - Z a z

1 a1030 -w Zaw

42

I1kb_jW W = =

-=° ° = 2 ° -0. la 6::Ev,XEu 0 10 *.C a

I t u IL 0ax nan-xmI.

IU-, I ) f

'phoA'I 50 R

ORI Apj

4--

x Q-x XI

I I Il i'tpse

15S50I

go I

Tn phoA

FIG. 4. Positions of TnphoA insertions in pPB1132. The cloned DNA in pPB1132 is indicated in panel a by the thick line. Each arrowindicates the position of the insertion of TnphoA, its orientation, and the resulting alkaline phosphatase phenotype. Closed arrowheadsindicate a positive alkaline phosphatase phenotype, and open arrowheads indicate a negative one. The structure ofTnphoA is shown in panelb, with orientation of the 'phoA fragment indicated. Numbers in brackets refer to acid phosphatase specific activities observed for strainCC118 isolates carrying different appA-phoA fusion plasmids. Boxed numbers refer to three fusions studied in detail.

95,,

67T

47p

40o

29.5 _w

o.7 ..

__ _b

40o

mAd*.

01

.-

A 04 004 - -A . . .

p- - MME04M MIF- - -L-

J. BACTERIOL.

2

L--t>I

on February 12, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

DETECTION OF GENES FOR EXPORTED PROTEINS IN E. COLI 1667

a95>

67o

47>

40t

22.5w

b.4

4 95o67T

.4

44>40o '

c

495

.l

'E

467

447

18.4A13.7Th-m

kDa 1 2 3

44029.5e.if':

.44 1 2 3 4 5 6 7 kDa

418.4

13.7i-u..-::.p

.i.

.O.,.....

kDa 1 2 3 4 5 6

FIG. 5. Molecular weight and stability of appA-phoA hybrid proteins. Cells of strain CC118 harboring plasmids pBR322, pPB1132,pPB1123, pPB1128, and pPB1127 were labeled with ['IS] methionine, and the corresponding lysates were immunoprecipitated with antiserumagainst acid phosphatase or against alkaline phosphatase. The figure shows an autoradiogram of a 11% polyacrylamide gel electropherogram.All samples correspond to an equal amount of cells. The positions of molecular weight standards are indicated. (a) Labeling for 3 min withoutchase. Lanes: 1, 2, and 3, CC118, respectively, with plasmids pPB1123, pPB1132, and pBR322, precipitated with antiserum against acidphosphatase; 4, CC118 with plasmid pPB1123 precipitated with antiserum against alkaline phosphatase.(b) Labeling for 1 min and immediatecell lysis. Lane 1, CC118(pPB1132) precipitated with anti-serum against acid phosphatase. Other lanes correspond to precipitation withantiserum against alkaline phosphatase of cells containing pBR322 (lane 2), pPB1132 (lane 3), pPB1128 (lane 4), pPB1123 (lane 5), andpPB1127 (lane 6). (c) Labeling for 2 min followed by a 20-min chase. All lysates were precipitated with serum against alkaline phosphatase.Lanes: 2, 4, and 6, respectively, pPB1128, pPB1123 and pPB1127 without chase; 3, 5, and 7, respectively, the same plasmids after the chase.

was precipitated by antibody to acid phosphatase from cellscarrying pPB1132. This band was faint in cells carryingpBR322 or no plasmid (Fig. 3b). These results show that thehigh level of acid phosphatase activity in cells with pPB1132is due to the overproduction of the enzyme.

Proteins encoded by plasmid pPB1132 were identified byusing the maxicell technique (20). The plasmid was found toencode a protein of the size expected for acid phosphatase(Fig. 3c). This protein was precipitated by antibody to acidphosphatase and was not encoded by pBR322. These resultsindicate that plasmid pBR1132 carries the appA gene.

Protein fusions between acid and alkaline phosphatasedefine the location and orientation of the appA gene. Where isthe appA gene located in plasmid pPB1132? To determinethis, random insertions of transposon TnphoA (16) intopPB1132 were isolated. If TnphoA inserted in the properorientation and translational reading frame into a geneencoding an exported protein, a hybrid protein with alkalinephosphatase activity could result. Thirty-three different in-sertions of TnphoA into pPB1132 were analyzed for alkalinephosphatase activity, acid phosphatase activity, and theposition of TnphoA insertion (Fig. 4). Of 19 alkaline phos-phatase-positive insertions, 9 completely eliminated acidphosphatase activity and mapped in a 0.8-kb region of the5.7-kb insert in pPB1132. In all such insertions, TnphoA wasin the same orientation. This region thus appeared likely tocontain the acid phosphatase structural gene.Nine other alkaline phosphatase-positive fusions showed a

decreased acid phosphatase activity (300 U/mg of protein,compared with 3,000 U/mg of protein in pPB1132) and

mapped in the tet region of the vector plasmid pBR322. Asimple interpretation of these insertions is that they generateactive tet-phoA fusions (16) and decrease the level of acidphosphatase by eliminating a contribution of the tet pro-moter to acid phosphatase expression in pPB1132. Indeed,when the 5.7-kb fragment of pPB1132 was present in theopposite orientation relative to vector sequences, so that thetet promoter could no longer contribute to appA expression(pPB1301), the level of acid phosphatase detected was alsoabout 300 U/mg protein. A similar difference in activity wasalso displayed by pPB1131 and pPB1318, in which a 6.8-kbfragment containing appA was cloned in the opposite orien-tation (Fig. 2). The remaining alkaline phosphatase positiveinsert into pPB1132 (star in Fig. 4) also showed a specificactivity of about 300 U/mg protein for acid phosphatase. Itsposition of insertion is between the tet gene and the putativeappA gene, so that it would also be expected to eliminate thecontribution of the tet promoter to appA expression inpPB1132. Since an insertion into this region gave an alkalinephosphatase-positive fusion plasmid, it is likely to encode aprotein exported from the cytoplasm (16). It is not knownwhy only one alkaline phosphatase fusion to this protein wasisolated.

Direct evidence for our assignment for the position andorientation of appA in the cloned DNA fragment came fromimmunological studies. A protein of 82 kilodaltons wasprecipitated from cells carrying plasmid pPB1123 (expectedto encode the longest appA-phoA fusion) by antibody toeither acid phosphatase or alkaline phosphatase (Fig. 5 a andc). From the size of appA-phoA fusion proteins (Fig. Sc), it

VOL. 169, 1987

40,14 I

I:

on February 12, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

1668 BOQUET ET AL.

TABLE 2. Characteristics of three appA-phoA fusions

Mol wt No. of residues Sp act (U/mg of protein)__________ __________ __ _ No. of residues Estimated size (kbp)

Plasmid Hybrid Acid phosphatase in hybrid or of remaining geneb Alkaline Acidprotein N-terminal fragment reference proteina phosphatase phosphatase

pPB1127 49,500 2,500 23 0.13 105 12.1pPB1128 53,000 6,000 54 0.22 63 10.9pPB1123 82,000 35,000 318 1.01 61 12.8pPB1132 No fusion 45,000 429 1.29 1 3,274

a Approximate number of acid phosphatase amino acid residues. Calculations were made by assuming that the acid phosphatase N-terminal amino acid in thehybrid was the same as that in the mature enzyme of wild-type strains (molecular weight of the alkaline phosphatase protein in TnphoA, 47,000; average molecularweight of one amino acid residue, 110).

b The estimated size of the remaining acid phosphatase gene, including a putative appA signal sequence of 0.06 kbp, was measured in an overnight culture inTYE medium with CC118 (appA +) as the recipient for the plasmids.

was possible to calculate that the appA gene starts about0.12 kb upstream from the site of TnphoA insertion inplasmid pPB1127. The appA gene would then extend about1.3 kb downstream from this position, since this is approx-imately the amount ofDNA required to encode a protein thesize of acid phosphatase (Table 2).The products of three fusion plasmids were further char-

acterized by immunoprecipitation. All three proteins showedinstability in pulse-chase experiments (Fig. 5c), with thelongest hybrid (encoded by pPB1123) showing an apparentdegradation product migrating at 46 kDa, the approximatesize of wild-type alkaline phosphatase. Similar stabilitybehavior was observed earlier for ,-lactamase-alkalinephosphatase hybrid proteins (16). In addition, for cellscarrying each of the three appA-phoA fusion plasmids, about85% of the alkaline phosphatase activity was released byosmotic shock (data not shown). This indicates the enzymat-ically active polypeptide was localized in the periplasm (18).This periplasmic location of alkaline phosphatase activityimplies that the signal promoting acid phosphatase export tothe periplasm can also function to promote alkaline phos-phatase export in hybrid proteins.

DISCUSSION

In this paper we describe the cloning of appA, the struc-tural gene for an E. coli acid phosphatase. The position andorientation of appA in a cloned DNA fragment were deter-mined by using fusions to alkaline phosphatase generated byTnphoA insertions. This fusion method appears to identifyselectively genes encoding periplasmic and membrane pro-teins (16) and should be generally useful in characterizing theproducts of cloned DNA fragments.The appA gene was cloned in vivo by using a derivative of

phage Mu (9) and then subcloned into the small multicopyplasmid pBR322. Cells carrying such recombinant plasmidsoverproduced acid phosphatase, and immunological analysisof plasmid proteins selectively labeled in maxicells indicatedthat the plasmids carried the appA gene. Further evidencesupporting this conclusion came from the finding that aderivative of one of these plasmids integrated in a region ofthe E. coli chromosome (22 min) known to contain the appAgene (unpublished re'sults).The enzyme produced by strains harboring one of the

recombinant plasmids (pPB1132) represents about 1% of thecellular protein and 60% of protein soluble in 1 M formicacid. This high level of production will aid the large-scalepreparation of this curious phosphatase for the purpose of itsstructural analysis.Mapping a gene on a recombinant plasmid is mostly

achieved by creating deletions and using insertions of

transposons (for a review, see reference 12). Both methodsinactivate, gene function but do not give information aboutthe orientation of the gene on the plasmid. The direction oftranscription may be obtained by operon fusions with lacZusing, for instance, spontaneous insertions of phage Mu dlinto the chromosome (24). Despite several attempts, suchfusions were not obtained with appA (Boquet, unpublisheddata). However, this gene, under usual conditions, is onlyexpressed in the stationary phase (5); thus, appA-lacZoperon fusions may not display a clearly detectable Lac'phenotype. In addition, hybrid proteins resulting from thefusion of lacZ to the N-terminal part of genes for secretedproteins usually show very little ,B-galactosidase activity (1).This might also explain the difficulty in obtaining lac fusions.Transposon TnphoA insertions in genes known to code for

proteins exported from the cytoplasm were previouslyshown to generate exported fusion proteins giving alkalinephosphatase activity (16). The results presented in this papershow that TnphoA insertions can be used to identify theposition and orientation of newly cloned genes for exportedproteins. Such an analysis of plasmids carrying appA (Table2) led to a map for the region containing appA (Fig. 6).A TnphoA insertion located approximately 300 base pairs

upstream of the unique ClaI site of appA does not abolishacid phosphatase transcription (Fig. 5). This result indicatesthat appA probably does not belong to an operon transcribedfrom an upstream promoter, but is preceeded by a promoterlocated within the 300-base-pair fragment to the left of theClaI site on pPB1132 (Fig. 6).DNA sequence analysis has confirmed that this region

1.29kb.

0 C4 OC-0

i . I

m O.Cf4CfO IzCnE In lu /M /t//2 -

. .-)CN to

2 _10 : ClILmLd

IA____iappA

1lkb , g

AC1C4 0

FIG. 6. Position and orientation of appA in a 4.6-kb restrictionfragment. The restriction map of the chromosomal region encodingappA showing the orientation and approximate position of the appAstructural gene. The positions of restriction sites and differentTnphoA insertion sites relative to the BamHI site are indicated.

___ _

J. BACTERIOL.

on February 12, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

DETECTION OF GENES FOR EXPORTED PROTEINS IN E. COLI 1669

encodes an open reading frame with an amino-terminalsequence resembling known signal sequences and that thissequence is preceeded by a typical -10 to -35 promoterregion and a Shine-Dalgarno sequence (Touati and Danchin,personal communication).The normal assay for acid phosphatase activity in colonies

on agar medium at low pH renders cells nonviable (4, 5).Replacing acid phosphatase activity by the alkaline phospha-tase activity of appA-phoA hybrid proteins makes it possibleto monitor appA expression at near-neutral pH by usingchromogenic substrates for alkaline phosphatase. This non-lethal assay should facilitate the isolation of mutants withaltered regulation of appA expression.

ACKNOWLEDGMENTS

We are grateful to Malcolm Casadaban for his gift of strainslysogenic for Mu cts and of plasmid pEG109 and to Jane Lopilato foradvice about the technique for maxicells. This work was done whileP.L.B. was on sabbatical in the laboratory of J.B., to whom he isparticularly indebted. The help of M. Hofnung and of P. Fromageotin facilitating the organization of this subbatical is gratefully ac-knowledged. We thank Ann McIntosh for excellent assistance inpreparing the manuscript.C.M. was a Fellow of the Arthritis Foundation. The work was

supported by an American Cancer Society Research Grant to J.B.and by the Commissariat A l'Energie Atomique, France.

LITERATURE CITED1. Bassford, P. J., Jr., T. S. Silhavy, and J. Beckwith. 1979. Use of

gene fusions to study secretion of maltose-binding protein intoEscherichia coli periplasm. J. Bacteriol. 139:19-31.

2. Bolivar, F., R. L. Rodriquez, P. J. Green, H. L. Betlach, H. L.Heynecker, H. W. Boyer, J. H. Crosa, and S. Falkow. 1977.Construction and characterization of new cloning vehicles. II. Amultipurpose cloning system. Gene 2:95-113.

3. Dassa, E., and P. L. Boquet. 1981. Is the acid phosphatase ofEscherichia coli with optimum pH 2.5 a polyphosphatedepolymerase? FEBS Lett. 135:148-150.

4. Dassa, E., and P. L. Boquet. 1985. Identification of the geneappA for the acid phosphatase (pH optimum 2.5) of Escherichiacoli. Mol. Gen. Genet. 200:68-73.

5. Dassa, E., M. Cahu, B. Desjoyaux-Cherel, and P. L. Boquet.1982. The acid phosphatase with optimum pH of 2.5 of Esche-richia coli, physiological and biochemical study. J. Biol. Chem.257:6669-6676.

6. Dassa, E., C. Tetu, and P. L. Boquet. 1980. Identification of theacid phosphatase (optimum pH 2.5) of Escherichia coli. FEBSLett. 113:275-278.

7. Davis, B. D., and E. S. Mingioli. 1950. Mutants of Escherichia

coli requiring methionine or vitamin B12. J. Bacteriol. 60:17-28.8. Dvorak, H. F., R. W. Brockman, and L. A. Heppel. 1967.

Purification and properties of two acid phosphatase fractionsisolated from shock fluid from Escherichia coli. Biochemistry6:1743-1751.

9. Groisman, E. A., B. A. Castilho, and M. Casadaban. 1984. Invivo DNA cloning and adjacent gene fusing with a mini-Mu-lacbacteriophage containing a plasmid replicon. Proc. Natl. Acad.Sci. USA 81:1480-1483.

10. Hafckensheid, J. C. M. 1968. Properties of an acid phosphatasein Escherichia coli. Bioch. Biophys. Acta 167:582-589.

11. Ito, K., P. Bassford, and J. Beckwith. 1981. Protein localizationin Escherichia coli: is there a common step in the secretion ofperiplasmic and outer membrane proteins? Cell 24:707-717.

12. Kleckner, N., J. Roth, and D. Botstein. 1977. Genetic engineer-ing in vivo using transposable drug-resistance elements. Newmethods in bacterial genetics. J. Mol. Biol. 116:125-159.

13. Laemmli, U. K. 1972. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

14. Mandel, M., and A. Higa. 1970. Calcium dependent bacterio-phage DNA infection. J. Mol. Biol. 53:159-162.

15. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

16. Manoil, C., and J. Beckwith. 1985. TnphoA: a transposon probefor protein export signals. Proc. Natl. Acad. Sci. USA82:8129-813.

17. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

18. Neu, H. C., and L. A. Heppel. 1965. The release of enzymesfrom Escherichia coli by osmotic shock and during the forma-tion of spheroplasts. J. Biol. Chem. 240:3685-3692.

19. Rogers, D., and F. J. Reithels. 1960. Acid phosphatases ofEscherichia coli. Arch. Biochem. Biophys. 89:97-104.

20. Sancar, A., A. M. Hack, and W. D. Rupp. 1979. Simple methodfor identification of plasmid-coded proteins. J. Bacteriol.137:692-696.

21. Schleif, R. F., and P. C. Wensink. 1981. Practical methods inmolecular biology. Springer-Verlag, New York.

22. von Hofsten, B. 1961. Acid phosphatase and the growth ofEscherichia coli. Biochim. Biophys. Acta 48:171-181.

23. von Hofsten, B., and J. Porath. 1962. Purification and propertiesof an acid phosphatase of Escherichia coli. Biochim. Biophys.Acta 64:1-12.

24. Wanner, B. L., S. Wieder, and R. McSharry. 1981. Use ofbacteriophage transposon Mud] to determine the orientation forthree proC-linked phosphate starvation inducible (psi) genes inEscherichia coli K12. J. Bacteriol. 146:93-101.

25. Wetzel, B. K., S. S. Spicer, H. F. Dvorak, and L. A. Heppel.1970. Cytochemical localization of certain phosphatases inEscherichia coli. J. Bacteriol. 104:529-542.

VOL. 169, 1987

on February 12, 2021 by guest

http://jb.asm.org/

Dow

nloaded from