Signal Peptide Subsegments Are Not Always Functionally ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY 8 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 264, No. 24, Issue of August 25. pp. 1447&14485.1989 Printed in U. S.A .

Signal Peptide Subsegments Are Not Always Functionally Interchangeable M13 PROCOAT HYDROPHOBIC CORE FAILS TOTRANSPORT

ALKALINE PHOSPHATASE IN

ESCHERICHIA COLI* (Received for publication, January 5, 1989)

Genevieve A. Laforet, Emil Thomas Kaiser?, and Debra A. KendallS From the Laboratory of Bioorganic Chemistry and Biochemistry, T h e Rockefeller Uniuersity, New York, New York10021

Bacterial signal peptides display little amino acid movement across the cytoplasmic membrane. Bacterial signal sequence homology despite their shared role in medi- peptides are characterized by a basic amino terminus, a cenating protein transport. This heterogeneitymay exist tral hydrophobic core, and a polar region preceding the cleavto permit the establishment of signal peptide conforage site (1).However, beyond these common features, signal mations that are appropriate for transport of particu- peptides show remarkably little amino acid sequence homollar proteins. In this paperwe explore how signal pep- ogy despite their shared role in mediating protein transport tides are composed of structural units that may interact (2). with each other and with the mature protein effect to Experiments suggesting that many random sequences can transport. Using a new application of cassette mutagenesis, we have replacedthe hydrophobic core of the serve as export signals (3, 4) and that signal peptides are functionally interchangeable (5) have been interpreted to Escherichia coli alkaline phosphatase signal peptide with cores from the signals of maltose-binding protein, mean that there are few functional constraints on the charOmpA, and M 1 3 major coat protein. The core regions acteristics of signal peptides. However, Inouye and co-workers from maltose-binding protein and OmpA effectively (6) have shown that identical mutant signal peptides display replaced the alkaline phosphatase core; the resultant widely different export efficiency depending on what protein hybrid signals performedas well as wild type in peri- they are transporting. Work by Kuhn et al. (7) demonstrated plasmic transport andprocessing of alkaline phospha- that reciprocal exchange of signal peptides between M13 coat tase. However, the core region from M13 major coat protein and OmpA results in impaired transport. Such eviprotein generated a transport-incompetent hybrid sig-dence suggests that individual signal peptide sequences are nal peptide. Elimination of a proline-containing por- optimized for transport of particular proteins. tion of the M 1 3 major coatprotein core did not improve Signal sequence heterogeneity also suggests the importance transport effectiveness. However, restoration of the of secondary structure over specific amino acid sequence procoat cleavage region and the negatively charged information in signal peptide function. Specifically, an aamino terminus of the mature protein did ameliorate helical conformationin the core region is thought to be the transportdefect. Theseresults suggest that at least required for transport (8);previous work from our laboratory in thecase of these procoat-derived signal peptide muhas demonstrated that a polyleucine core region with a high tants,there is a requirementforcomplementarity propensity for hydrophobic a-helix formation enhances transamong the hydrophobic core, cleavage region, and part port efficiency (9). Similarly, the ability of isolated synthetic of the mature protein in order for efficient protein signal peptides to form certain secondary structures correlates transport to occur. with their transport competence i n uiuo (10). Despite these observations, a precise understanding of how signal peptide secondary structure mediates proteintransport Many prokaryotic proteins thatare synthesized in the has remained elusive. Most attempts to study the structural cytoplasm ultimately reside in noncytoplasmic compartments. requirements for signal peptide function have focused on To achieve their final localization, most transported proteins single amino acid deletions or substitutions that impair or require an amino-terminal signal peptide to facilitate their restore transport (5, 11).In this paper we take a more global approach, treating the components of the signal peptide* This research was supported in part by National Institutes of especially the hydrophobic core and cleavage region-as whole Health Grant GM37639 (to D. A. K.) and by the R. J. Reynolds Fund structural units to begin to understand how these segments for the Biomedical Sciences and Clinical Research, Winston-Salem, NC (to G. A. L.). The costs of publication of this article were defrayed may interact structurally and functionally to mediate protein in part by the payment of page charges. This article must therefore transport. be hereby marked “aduertisement” in accordance with 18 U.S.C. Using a new application of cassette mutagenesis to the Section 1734 solely to indicate this fact. study of signal peptides (12), we have inserted different natDedicated to the memory of Emil Thomas Kaiser, who created a vibrant and cooperative scientific environment without which this ural core regions into the signal peptide of the Escherichia coli periplasmic enzyme alkaline phosphatase in place of the work would not have been possible. T h e nucleotide sequence(s)reported in this paperhas been submitted native sequence. Such an approach permits the comparison accession number(s) of the transport competence of different core regions operattotheGenBankTM/EMBL Data Bankwith 505005. ing in the same amino acid framework and directing export t Deceased July 18, 1988. $ To whom correspondence should be addressed Dept. of Molecu- of the same protein. With this approach, we have demonlar and Cell Biology, the University of Connecticut, Box U-44, 75 N. strated that individual core regions, although all functional in their nativeenvironment,can differ in their ability to Eagleville Rd., Storrs, CT 06269.

14478

Signal Peptide Hydrophobic

Cores Are Not Always Interchangeable

direct transportwhenplacedin the alkaline phosphatase signal. Although most mutants approximated wild-type efficiency, one was virtually transport-incompetent.In the nonfunctional mutant, transport function can be improved by restoring the matching cleavage sequencenext to the core and further enhanced by placing a small negatively charged portion of the mature protein next to the core and cleavage regions. This suggests that at least in some cases there is an interaction among the core, cleavage region, and mature protein which optimizes transport of a given protein. In other words, signal peptide function may be determined by global or overall properties that can be imparted only bythe sum of these segments. To generate the mutants for these studies, we selected core regionsfrom the signals of three representativeproteins: an maltose-binding protein, a periplasmic protein; OmpA, outer membrane protein; and M13 major coat protein, an inner membrane protein. These core regions were chosen for comparison because they are all the same length yet have widely different amino acid compositions and derive from proteins that are targeted to different cellular compartments. Fig. lA shows the amino acid sequences and distinguishing features of each core region chosen. The core was defined as beginning afterthe last polar or charged residue of the aminoterminal regionand ending afterthe last cluster of hydrophobic amino acids inthe signal sequence (1). MATERIALS ANDMETHODS

Bacterial Strains The strain used in this paper is E. coli AW1043 (Alac galU galK A(leu-ara)phoA-E15 proC::Tn5). This strain contains a partially deleted alkaline phosphatase gene and a wild-typephoR regulatory gene (13, 14). Media Bacteria were propagated in LB medium or plates containing 50 pg/ml kanamycin and 250 pg/ml ampicillin (15). For transport studies, cells were cultured in M O P S medium (16) with antibiotics as above, under low phosphate conditions (100 pM) or zero phosphate to induce alkaline phosphatase expression. Construction and Expression of Signal Peptide Mutants Construction of our mutant signal peptides utilized a cassette mutagenesis vector containing the alkaline phosphatase structural gene phoA in which unique Sal1 and BssHII sites bracket the hydrophobic core coding region (12). Removal of the wild-type core DNA using these enzymes permitted insertionof synthetic oligonucleotides encoding the new core regions. Sequence information used to generate these oligonucleotides was obtained from Refs. 17-19. The DNA and amino acid sequences of the resultant constructs are shown in Fig. 1B. Fidelity of the insertions was verified by restriction analysis and direct DNA sequencing (20). The resultant plasmids were used to transform E. coli strain AW1043 (12). Mutant hcmPC-AP (Fig. 1B) was constructed by BssHII cleavage of mutant plasmid hcPC-AP (Fig. 1B) followed by treatment with S1 nuclease as follows. After digestion with BssHII (New England BioLabs), hcPC-AP was extracted with phenol-chloroform, ethanolprecipitated, and then incubated with serial dilutions of S1 nuclease (Boehringer Mannheim). The generation of blunt ends was verified by increased mobility on SDS-polyacrylamide gels (15). After S1 nuclease treatment, the vector was phenol-chloroform-extracted, ethanol-precipitated, and then used for cassette mutagenesis as described (12).

14479

Cells were harvested in the logarithmic phase of growth, washed, and resuspended in MOPS with zero phosphate supplemented with amino acids at 2 pg/ml minus methionine. After resuspension, cells were further diluted 1 : l O in this same medium prior to labeling with 42 pCi of [35S]methioninefor 90 s. For whole cell samples, labeling was stopped by addition to an equal volume of ice-cold 10% trichloroacetic acid. Precipitates were washed twice in cold acetone and air-dried. Protein was resolubilized by boiling for 3 min in 10 mM Tris, pH 8, 1%SDS, 1 mM EDTA then diluted by the addition of 50 mM Tris, pH 8,150 mM NaC1,O.l mM EDTA, 2% Triton X-100. For periplasmic samples, incorporation of label was stopped by rapid cooling on ice. Aliquots of labeled culture were then washed in 30 mM Tris, pH 8, and resuspended in 0.5 M sucrose, 30 mM Tris, pH 8, containing 1 mM EDTA and 20 pg/ml freshly dissolved lysozyme (Sigma). After incubation at room temperature for 20 min, samples were centrifuged, and the supernatants containing the periplasmic fraction were passed through a 0.22-pm filter prior to dilution with 50 mM Tris, pH 8,150 mM NaC1, 2% Triton X-100, 0.1 mM EDTA. Whole cell and periplasmic samples were then subjected to immunoprecipitation using (12) essentially 2.5-5 pl ofrabbit anti-alkaline phosphatase antiserum according to the method of Ito et al. (21). Pulse-Chase Analysis-Bacteria were subcultured, washed, and resuspended as described above. In certain cases, to enhance incorporation of radioactivity, cells were further diluted 1:lO prior to labeling as outlined in the periplasmic cell fractionation protocol. Cells were labeled with 42 pCi of [35S]methioninefor 40 s at 37 "C then chased with 4 mg/ml nonradioactive methionine for 30 s, 1 min, 5 min, and 15 rnin. The chase was terminated by the addition of an equal volume of ice-cold 10% trichloroacetic acid. The resultant precipitates were washed with cold acetone, dried, and boiled in 10 mM Tris, pH 8, 1%SDS, 1 mM EDTA. Resolubilized precipitates were then diluted in 50 mM Tris, pH 8, 150 mM NaC1, 2% Triton X100, 0.1 mM EDTA prior to immunoprecipitation using rabbit antialkaline phosphatase antiserum as described above. Sodium Hydroxide Cell Fractionation-This cell fractionation protocol followed the method of Russel and Model (22). Cells weretreated exactly as inthe periplasmic cell fractionation above except that cells were not diluted prior to labeling, and incorporation of label was stopped by the addition of an equal volume of ice-cold freshly diluted 0.2 N NaOH. Samples were centrifuged for 15 min at 4 "C to separate the NaOH-stripped membranes from soluble components. Both the membrane and soluble fractions were precipitated in cold 10% trichloroacetic acid. Precipitates were processed exactly as inthe pulsechase protocol above. SDS-Polyacylamide Gel Electrophoresis and AutoradiographyImmunoprecipitated alkaline phosphatase was run on Laemmli SDSpolyacrylamide gels (23) and subjected to autoradiography asdescribed (12). Amino-terminal Radiosequencing by Automated Edman Degradation Alkaline phosphatase from the signal peptide mutants was radiolabeled and isolated for Edman degradation (24) essentially as in (12) except for the following variations. In hPC-AP and h'PC-AP, incorporation of label was quenched by the addition of an equal volume of 100% trifluoroacetic acid to ensure that the amino-terminal methionine of the alkaline phosphatase precursor was deformylated. After 1 h on ice, protein was precipitated in cold 10% trichloroacetic acid. Precipitates were acetone-washed, dried, and resolubilized as in the pulse-chase protocol above prior to SDS-polyacrylamide gel electrophoresis and transfer to Immobilon (Millipore) (25) as described in (12). In the case of mutant hcmPC-AP, the amino-terminal sequence was determined by high pressure liquid chromatography identification of phenylthiohydantoin derivatives obtained by automated Edman degradation. RESULTS

The ability of mutants hMBP-AP, hOA-AP, and hPC-AP (Fig. lA)to export alkaline phosphatase to the E. coli periplasm was tested by lysozyme-EDTA cell fractionation. As Transport Studies shown in Fig. 2, the wild-type alkaline phosphatase signal is Periplasmic Lysozyme-EDTA Cell Fractionation-E. coli cultures able to transport the mature enzyme quicklyand completely harboring mutant plasmids were growna t 37 "C in MOPS containing 100 p M phosphate, 50 pg/ml kanamycin, and 250 pg/ml ampicillin. to the periplasmic space. The core region mutants, however,

The abbreviations used are: MOPS, 4-morpholinepropanesulfonic acid SDS, sodium dodecyl sulfate.

vary in their ability to transport alkaline phosphatase. M u t a n t hMBP-AP functions well, generating periplasmic levels of mature alkalinephosphatase comparable to wild type. Mutant

Signal Peptide Hydrophobic Cores Are Not Always Interchangeable Derivation of Inserted Sequence

Location of Mature Protein

Amino Acid Sequence of Signal Peptide

Special Features

Wild Type

alkaline phosphatase

periplasm

MKQSTIALALIZLU'TPVTKA'

wild type alkaline phosphatase

hMBP-AP

maltose-binding protein

periplasm

MKQSTILALSALTTMMFTPVTKA'

core region + polar amino acids + methionines

hOA-AP

Omp A

outer membrane

MKQSTAIAIAVALAGFATPVTKA'

core region alanine-rich leucine-poor

hPC-AP

M13 major coat protein

inner membrane

MKQSTASVAVATLVPMLTPVTKA'

core region valine-rich

h'PC-AP

M 13 major coat protein

inner membrane

MKQSTASVAVAmVTPVTKA'

shorter M13 procoat core region

hCPC-AP

M 13 major coat protein

inner membrane

MKQSTASVAVATLVPMLSFA'

hydro hobic core and &wage region from M13 procoat

hcmPC-AP

M13 major coat protein

inner membrane

MKQSTASVAVATLVPMLSFA'AEGDDPA

hydrophobic core, cleavage region and first 7 amino acids of M13 procoat

hPC-AP K-F

M13major coat protein

inner membrane

MKQSTASVAVATLVPMLTPVTFA'

M13 core region lys "t phe substitution at -2

hPC-AP K-Q

M13 major coat protein

Inner membrane

MKQSTASVAVATLVPMI.,TPVTQA'

M13 core region lys "L gln substitution at -2

A

Mutant

"

B

Mum WT CASS 3

hMBP-AP

Met Lys G l n S e r T h r Ile Ala Leu Ala Leu Leu Pro Leu AAA CAG TCG ACT ATT GCA CTG GCA CTC TTA CCG TTA

+ GTG

leuPhe T h r P r o V a l T h r Lys Ala'Arg CTG TTT ACC CCTGTG ACA AAA GCGCGC

Thr ACA

G l n S e r T h r Ile leu Ala leu S e r Ala leu Thr Thr Met Met Phe T h r P r o V a l T h r Lys Ala'Arg ATC CTC GCA TTA TCC GCA TTA ACG ACG ATG ATG TTT ACC CCT GTG ACA AAA GCGCGC

Thr ACA

Met Lys G l n S e r T h r Ala Ile Ala Ile Ala V a l Ala Leu Ala G l y Phe Ala T h r P r o V a l T h r Lys Ala'Arg AAA CAG TCG ACT GCT ATC GCG ATT GCA GTG GCA CTG GCT GGT T T C GCT ACC CCT GTG ACA AAA GCGCGC

Thr ACA

Met

+ GTG Lys CAG TCG ACT AAA

hOA-AP

+ GTG

hPC-AP

+

Met Lys G l n S e r T h r Ala Ser V a l Ala V a l Ala Thr Tau V a l Pro Met Leu T h r Pro V a l T h r Lys A l d A r g T h r GTG AAA CAG TCG ACT GCC TCC GTA GCC GTT GCT ACC Cn: GTT CCG ATG CTG ACC CCT GTG ACA AAA GCGCGC ACA

t

Thr Met Lys G l n S e x T h r Ala Ser V a l Ala V a l Ala Thr Tau V a l T h r P r o V a l T h r Lys Ala'Arg GTG AAA CAG TCG ACT GCC TCT GTA GCC GTT GCT AW CTC GTT ACC CCT GTG ACA AAA GCG CGC ACA

h'PC-AP

Thr e t Tau Ser Phe Ala'Arg Met Lys G l n S e r T h r Ala Ser V a l Ala V a l Ala Thr Tau V a l Pro W AAA CAG TCG ACT GCC TCT GTA GCC GTT GCT ACC CTC GTT CCG ATG CTG TCT TTC GCG CGC ACA

hcPC-AP

+ GTG

hcmPC-AP

+

hPC-AP P F

+ GTG AAA CAGTCGACT

hPC-AP lt9 t

Met Lys G l n S e r T h r Ala Ser V a l Ala V a l Ala Thr Leu V a l Pro Met Tau Ser Pha Ala'Ala G l u Gly AspAap Pro A l a A r g T h r GTG AAA CAGTCGACT GCC TCT GTA GCC GTT GCT CTC ACC GTT CXG ATG CTG TCT TTC GCT GCT GAG GOT GAC GAT CCC GCA CGC ACA Met Lys G l n S e r T h r Ala Ser V a l Ala V a l Ala Thr leu V a l Pro W e t leu T h r Pro V a l T h r P h e Ala'Arg GCC TCT GTA GCC GTT GCTACC CTC GTTCCG ATG CTG ACC CCT GTG ACA TTCGCGCGC

Thr ACA

Met Lys G l n S e x T h r Ala Ser V a l Ala V a l Ala Thr Leu V a l Pro Met Leu T h r Pro Val T h r G l n Ala'Arg GTG AAA CAG TCG ACT GCC TCT GTA GCC GTT GCT CTC ACC GTT CCG ATC CTG ACC CCT GTG ACA CAG GCGCGC

Thr ACA

FIG. 1. A, characteristics of the sequences used to generate mutant hybrid signal peptides. Definition of core region boundaries followed the principles outlined by von Heijne (1). The mutant aminoacid sequences are shown in boldface. An apostrophe marks the site of cleavage by signal peptidase. B, amino acid and DNA sequences of mutant signal peptides. W T CASS 3 contains unique Sal1 and BssHII sites that were introduced into the wildtypealkalinephosphatase signal peptidewithoutchanging the nativeamino acid sequence (12). All results pertaining to "wild type' in this paper refer to WTCASS 3. This vector was used to generate mutants hMBP-AP, hOA-AP, hPC-AP, h'PC-AP, hcPC-AP, hcmPC-AP, hPC-AP K-F, and hPC-AP K-Q by cassette mutagenesis as described in (12).Amino acid and nucleotide sequences from M 1 3 major coat protein, maltose-binding protein, and OmpA were obtained from Refs. 17-19, respectively. For clarity, only the + strand of the DNA sequence has been shown. Nucleotides that differ from wild-type sequences are italicized. In mutant hPC-AP, the nucleotide sequence from phage fd rather than from M 1 3 was used the only difference is a T to C change in the wobble position of codon 7 (serine). The amino acid and nucleotide sequences of the substituted regions are shown in boldface. An arrowhead marks the siteof cleavage by signal peptidase.

Signal Peptide Hydrophobic Cores Are Not Always Interchangeable hPC-AP

-

hMBP-AP

14481

hOA-AP

WT

" " 30" I' 5' 15' 30" I' 5' 15' 30" I' 5' 15' 30" I' 5' 15'

0-

bD

a-

ea

1

b*

O r ) - ( l ) - a b

FIG.2. Lysozyme-EDTA cell fractionation of core region mutants. Bacteria expressingwild-type ( W T ) or mutant signal peptide core regions were radiolabeled, and protein from periplasm and whole cells wasprepared asdescribed under "Materials and Methods." Arrowheads mark the positionsof precursor ( a ) and mature( b )forms of alkaline phosphatase in unfractionated whole cells ( WC)uersus isolated periplasmic fractions ( p e r i ) .

" "

S O D

bD

P

S

"

I

P .

.

S I

"

P

~

S .

.. P 40

-

0

r b

FIG.3. Sodium hydroxide cell fractionation of core region mutants. Radiolabeled cells werefractionated intosoluble and membrane components using NaOH as described under "Materials and Methods." Arrowheads indicate the positions of precursor ( a ) and mature ( b ) forms of alkaline phosphatase in thesoluble fraction (S) consisting of cytoplasm and periplasm, or the pellet ( P ) containing membrane-associated proteins. WT, wild type. hOA-AP also transports all the alkaline phosphatase synthesized to the periplasmic compartment. In contrast, the periplasmic fraction from mutant hPC-AP reveals no detectable alkaline phosphatase. The hybrid signal that functioned most poorly in transport to the periplasm was derived from M13 major coat protein, an inner membrane protein. Although the key determinants for protein targeting in bacteria are thought to rest primarily with the mature protein (26), a recent theoretical analysis suggests that some of the information for localization is contained in the physical characteristics of the signal peptide (27). Therefore, each signal peptide mutant was tested for localization to membranes (inner and outer) or soluble compartments (cytoplasm and periplasm) using asodium hydroxide cell fractionation method (22). Fig. 3 shows the results from this analysis. Mutants hMBP-AP, hOA-AP, and wild type all show only mature enzyme located predominantly in the soluble fraction (Fig. 3). This observation is consistent with the results from the lysozyme-EDTA cell fractionation, indicating rapid and complete localization of processed enzyme to the periplasm (Fig. 2). In contrast, mutant hPC-AP againfunctionsdifferently,witha substantialamount of unprocessed enzyme partitioning with the membrane. Not all of the protein is associated with the membrane, however, suggesting that this behavior may simply reflect abortive transport rather thana true membrane localization event (5, 28). The doublets observed in Fig. 3 occur commonly with the sodium hydroxide method and apparentlyreflect slightly different migration of the sameprotein.2 Because transport to theperiplasmic side of the membrane is a prerequisite for cleavage by signal peptidase(29), impaired

* N. Davis and M. Russel, personal

communication.

I

"1

40

-b

FIG. 4. Pulse-chase analysis of core region mutants. Radiolabeled cellswere subjected to pulse-chase analysis asdescribed under "Materials and Methods." Arroloheads mark the positionsof precursor ( a ) and mature ( b )forms of alkaline phosphatase. WT, wild type.

export is usually accompanied by reduced efficiency of precursor processing. As shown in pulse-chase analysis inFig. 4, cleavage of the wild-type signal peptide is rapid, with only trace amountsof precursor discernible at thetwo earliest time points. In contrast, the core region mutants exhibit marked differences in processing efficiency. Again, mutant hMBP-AP is comparable to wild type, with only low levels of precursor evident a t 30 s and 1 min, after which time only mature enzyme can be detected. Similarly, hOA-AP shows extremely rapid processing with virtually no precursor detectable even at the earliest time point. In contrast with these findings, however, hPC-AP again shows a severe defect: within the limits of detection of this experiment, there is absolutely no processing to the matureform even at the15-min time point. To document the processing behavior of our mutants, we performed amino-terminal radiosequencing to verify their site of cleavage by signal peptidase. As expected, untransported mutant hPC-AP shows no cleavage by signal peptidase (Fig. 5 A ) , whereas periplasmic alkaline phosphatase from mutant hMBP-AP shows faithful cleavage at the normal processing site (Fig. 5 B ) .The twin peaks of radioactivity a t cycles 4 and 5 reflect formation of an isozymic variant of alkaline phosphatase (30) by the E. coli iap protease, which selectively removes the amino-terminal arginine of the mature enzyme (31). This protease can be inhibited by the presence of 10 mM arginine or by the protease inhibitor leupeptin (32). When cells were grown and labeled in the presence of these inhibitors, isozyme formation was blocked, but correct signal peptidase cleavage specificity was maintained (data not shown). These radiosequencing results indicate that replacement of the core region can alter theefficiency of processing but does not change the specificity of the cleavage event when it does occur. These experiments show that although hydrophobic core regions from maltose-binding protein and OmpA can functionally replace the core region from alkaline phosphatase, the core from M13 procoat cannot. There are a number of possible mechanistic or structural bases for this failure. The simplest possibility is that the procoat core region has been defined inappropriately (33). Alternatively, there could be some structural requirement for complementarity between the procoat core and cleavage region which is violated by isolating just thecore region within the alkaline phosphatase signal. Also, procoat transport is thought to require the presence of a number of negative charges in the amino terminus of the mature protein (29). The procoat signal may not be able to mediate transport of alkaline phosphatase in the absence of this charged region. To test these alternatives, we constructed three mutants with the following characteristics (Fig. 1, A and B ) . Mutant h'PC-AP contains a 9 residue procoat core that eliminates the final Pro-Met-Leu present in the hPC-AP core. This shorter core region omits a highly unusual second proline residue that might cause transport-impairing structural perturbations in the hybrid signal. Mutant hcPC-AP contains the procoat cleavage region in addition to the core. Mutant

Signal Peptide Hydrophobic Cores Are Not Always Interchangeable

14482

A

50000l

40WO

t: ’OWO

b

2oaoo

-

-

;

zowo

-

low0

-

loo00

I

~

2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 l 9 2 0 2 l 2 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9 3 0 CYCLE#

H ~ O S T A S V A V A l L V P H L T P V T K A R T P E H P V P E l l D M

I

2

3

4

5

6

7

8

9

10

I1

12

R

T

P

E

n

P

v

L

E

N

R

A

T

P

P

H

P

V

L

E

N

A

A

n

R

CYCLE) RESIDUE

n

H K O S T A S V A V A T L V T P V T K A n R T P E H P V L E N R R E S I D U L

R

T

P

E

T

P

E

H

H P

P

v

1

V

L

E

E

N

N

R

RESIDUE

A

FIG. 5. Radiosequencing of signal peptide mutants. Cells bearing mutant plasmids were labeled with [35S] methionine and their alkaline phosphatase isolated as described under “Materials and Methods.” The bar graph represents peaks of radioactivity determined for each cycle of Edman degradation. The naturalamino acid sequence correlating with the experimental results is shown under each plot. Panel A, radiosequencing of hPC-AP. Peaks of radioactivity at cycles 1 and 16 correspond to the sequence of unprocessed mutant precursor. Panel B, radiosequencing of hMBP-AP. Peaks of radioactivity at cycles 4 and 5 correspond to thesequence of faithfully processed mature alkaline phosphatase and its isozymic variant from which the amino-terminal arginine has been removed. Panel C, radiosequencing of h’PC-AP. Peaks of radioactivity at cycles 1 and 25 correspond to the sequence of unprocessed mutant precursor. Panel D, radiosequencing of hcPC-AP. Peaks of radioactivity at cycles 4 and 5 correspond to thesequence of correctly processed mature alkaline phosphatase and its isozyme.

hcmPC-AP builds onto hcPC-AP by adding the first 7 amino acids of mature coat protein, including three critical negatively charged residues. Figs. 6 and 7 show that redefinition of the core region in mutant h‘PC-AP did not increase the transporteffectiveness of the M13 hybrid. There is still no detectable periplasmic enzyme, nor is there any processing to the mature form over the time course examined. Amino-terminal radiosequencing is consistent with the presence of the precursor form only (Fig. 5 C ) . However, restoration of the cleavage region next to the core results in some improvement of the transport competence of the procoat-derived signal. Although not reaching levels comparable to wild type, mutant hcPC-AP does accumulate modest amounts of mature alkaline phosphatase in the periplasm (Fig. 6), with slow and incomplete but nevertheless detectable processing of alkaline phosphatase precursor to the mature form (Fig. 7A). Amino-terminal radiosequencing verifies that this mature form is processed at the correct cleavage site (Fig. 50). Finally, addition of the nega-

tively charged coat protein amino terminusmutant in hcmPCAP further enhances efficiency of transport (Fig. 6) and processing (Fig. 7A) over hcPC-AP, which contained simply the core and cleavage regions. This mutant, although processed, migrates like alkaline phosphatase precursor (a) because it contains an extra 7-residue negatively charged segment from coat protein. Amino-terminalsequencing confirms that this periplasmic material is the expected mature coat protein-alkaline phosphatase hybrid, processed at thecorrect cleavage site (Fig. lB,data not shown). The improvement in transport and processing in our procoat-derived mutants upon addition of negative charges to the mature amino terminus supports earlier findings that a negative charge balance around the cleavage region is necessary for efficient cleavage by signal peptidase (34). It is possible that the progressive enhancement of transport function observed upon sequential restoration of procoat-derived segments toour alkaline phosphatasehybrid could be due merely to anincrease in the negativity of the region surrounding the

Signal Peptide Hydrophobic Cores Are N o t Always Interchangeable

-

" "

b-

m-

I

, ( I l c n

-0

-a cb

Y

FIG.6. Lysozyme-EDTA cell fractionation of procoat-derived signalpeptide mutants. Cell fractionation was performed as described under "Materials and Methods." Arrowheads mark the positions of precursor ( a ) and mature ( b ) forms of alkaline phosphatase in whole cell ( W C ) and periplasmic (peri) fractions. In the case of mutant hcmPC-AP, the material migrating at position ( a ) actually corresponds not to uncleaved precursor but to a processed hybrid protein containing the first 7 amino acids of M13 major coat protein fused to the mature portion of alkaline phosphatase (see "Results"). WT,wild type.

14483

The transportproperties of these mutantswere first tested by pulse-chase analysis as shown in Fig. 7B. Both mutants K-F and K+Q displayincreased processing of alkaline phosphatase over their nonfunctional parent mutant hPCAP. However, processing efficiency is still below that observed for wild-type alkaline phosphatase. Comparison of Fig. 7B, hPC-AP K+F and hPC-AP K+Q, with Fig. 7A, hcPC-AP, demonstrates that the degree of improvement observed in K-F and K-Q approximates that seen in mutant hcPC-AP, which restored the entireprocoat cleavage region next to the core (Fig. lA). Similarly, lysozyme-EDTA cell fractionation also revealed slightly improved periplasmic transport of mature enzyme in these mutants compared with hPC-AP, but again these levels were significantly lower than those seen in wild type (data notshown). Thus, simple removal of a single positive charge a t position -2 from mutant hPC-AP results in an improvement of transport and processing efficiency equivalent to thatseen with restoration of the entireprocoat cleavage region in mutant hcPC-AP. DISCUSSION

30" I' 5' 15"30" I' 5' 15'30'' I' 5' 15"30" I' 5' 15" ,.

^I_

-a

#

"

30" I' 5' 15' 30" I' 15' 5'

30" I' 5' 15'30" I' 5' 15'

l

a, b-

-a -b

FIG. 7. Pulse-chase analysis of procoat-derived signal peptide mutants. Pulse-chase experiments were performed as described under "Materials and Methods." Chase times are indicated above each l a n e . Arrowheads mark the positionsof precursor ( a ) and mature ( b ) forms of alkaline phosphatase (see "Results" and the legend to Fig. 6 concerning slow migration of processed form of mutant hcmPCAP). A, mutants h'PC-AP, hcPC-AP, hcmPC-AP, and WT CASS 3, ( WT).B, mutants hPC-AP, hPC-AP K-F, hPC-AP K-Q, and WT CASS 3 ( W T ) .

cleavage site, first by elimination of a positively charged lysine a t position -2 (hcPC-AP, Fig. LA) then by further addition of three negative charges tothemature amino terminus possibility, we constructed (hcmPC-AP, Fig. lA).To test this two additional mutants, hPC-AP K+F and hPC-AP K+Q, in which the lysine a t position -2 of the transport-incompetent procoat-derived mutant hPC-AP was replaced by either phenylalanine or glutamine (Fig. 1, A and B ) . Phenylalanine was chosen because it is the most common residue occurring a t position -2 of prokaryotic signal peptides(35). Glutamine was chosen because it is uncharged but morepolar than phenylalanine and thusmore closely approximates the hydrophilicity of the native alkaline phosphatase cleavage region. Also, empirical studies have suggested a correlation between propensity for @-turnformation around the cleavage region and efficiency of precursor processing (36);glutamine iscomparable to lysine in probability for turn formation according to Chou-Fasman parameters, whereas phenylalanine has a much lower probability for turn formation (37).

Our results show that there can be strong constraints on the sequence characteristics of signal peptides in order for effective transport andprocessing to occur. By systematically dissecting the structuralcomponents of individual signals, we can demonstratedifferences in the ability of different natural hydrophobic core sequences to direct transport when placed in the samecontext, the alkaline phosphatase signal peptide. Although in the case of hMBP-AP and hOA-AP, wild-type transport and processing efficiency are maintained,in hPCAP, export and processing are abolished. This defect can be almost totally overcome by restoration of the cleavage region and part of the mature portion of M13 major coat protein, suggesting that there is an optimalcombination of these segments which interacts structurally and/or functionally to mediate transport. By treating segments of the signal peptide as whole units, the approach taken here helps refine our understanding of how global signal peptide features govern transport function. This approach was facilitated by the use of cassette mutagenesis, which allows the precise manipulation of target domains withoutintroduction of adventitiouschangesinadjoining sequences. Such a system permits clear comparisons of the transport competence of different combinationsof signal peptide subsegments in the samewild-type alkaline phosphatase context. The amino acid sequences chosen all derived from proteins normally transported in E. coli, and wild-type codon usage was preserved throughout. All our mutants were expressed in the same multicopy plasmidsystem t o permit controlled side-by-side evaluations of transport efficiency. The evidence from these signal peptide mutants suggests t h a t a tleast in the case of our procoat-alkaline phosphatase hybrid, there is a requirement for complementarity among the hydrophobic core, cleavage region, and part of the mature protein in order for efficient protein transport to occur. To understand the mechanistic basis for this requirement, it is logical to consider first the physical and structural characteristics of the mutantsignals themselves. One critical determinant of transport function is signal peptide hydrophobicity (10,38). By rearranging combinations of signal peptide segments, it is possible that we altered the nethydrophobicity of the signal below a threshold value required for transport to occur. However, in analyzing our mutant sequences, there were no patternsin the hydropathy profiles (39)of either the mutant signals or various combinations of their subsegments which could account for the observed differences in transport

14484

Signal Peptide Hydrophobic Cores Are Not Always Interchangeable

effectiveness (data not shown). Also, signal peptide confor- phatase hybrid containing a negatively charged sequence inmation is thought to be important for the membrane associ- volved in sec-independent transport invites the speculation ation and insertion steps of export. In particular, studies on that the sec-independent export pathway may be usable by synthetic signal peptides suggest that a transition between a normally sec-dependent proteins if the proper structural ele@-structure and an a-helix insignal the may be central to this ments for use of that pathway are present. Future experiments using various mutant sec strains will allow definitive examiprocess (10). Interactions between adjacent structural units in the signal and mature proteinmay help effect the required nation of this issue. In summary, our results indicate that for export of some conformational transitions to facilitate insertionin vivo (40). However, we have no evidence for differences in conforma- proteins, components of the signal peptide such as the core tional preference between our functional and nonfunctional and cleavage regions must complement one another and the peptides. Notwithstanding the limited applicability of existing mature protein in order tomediate transport successfully. In structural predictive schemes to membrane-interactive pro- particular, additionof negative charges in the amino terminus teins (41), Chou-Fasman analysis (37) of our mutant signals of the mature protein can compensate for a mutant signal revealed no correlation between transport competence and peptide with suboptimal transport properties. Further applipropensity for formation of any particularsecondary structure cation of cassette mutagenesis to the amino terminus, core, cleavage region, and mature protein should further illuminate (data not shown). these Thus, we candiscernno obviousphysical or structural the structural and functional interactions amongregions property that would account for the requirement for comple- which govern protein transport. mentarity amongcore, cleavage region, and mature protein in Acknowledgments-The superb technical assistance of Mark Bianour procoat hybrids. However, we cannot rule outmore complex mechanistic reasons for this requirement. Signal peptides chi, Margaret Chou, and Suzanne Doud is gratefully acknowledged. We wish to thank PeterModel and Lila Gierasch for helpful discusare implicated in the maintenanceof proteins in a transport- sions and Stephen Benkovic for comments on this manuscript. Procompetent conformationby delaying their folding into a tight tein sequence analysis was performed courtesy of Donna Atherton at tertiary structure (42, 43). Particular combinations of signal The Rockefeller University Protein Sequence Facility, which is suppeptide segments tailored to a given mature protein may be ported in part by funds provided by the United States Army Research necessary to optimize this folding effect. The signal peptide office for the purchase of equipment. is also thought to interact with components of the transport REFERENCES machinery including signal peptidase, trigger factor (44, 45), 1. von Heijne, G. (1985) J. Mol. Biol. 1 8 4 , 99-105 and perhaps some of the sec gene products (10). Unlike the other proteins examined in this paper, M13 major coat protein 2. Watson, M. E. E. (1984) Nucleic Acids Res. 12, 5145-5164 C. A., Preuss, D., Grisafi, P., and Botstein, D. (1987) does not require any of the sec-encoded products for its correct 3. Kaiser, Science 235,312-317 maturation andassembly into membranes (29). Although the 4. Smith, H., Bron, S., van Ee. J., and Venema, G. (1987) J. determinant for procoat's independence of secY residesin the Bacteriol. 169, 3321-3328 5. Benson, S. A,, Hall, M. N., and Silhavy, T. J. (1985) Annu. Reu. mature protein as opposed to the signal itself ( 7 ) , a procoatBiochern. 54, 101-134 derivedsignal may be unabletofunctionintransport of 6. Lehnhardt, S., Pollitt, S., and Inouye, M. (1987) J. Biol. Chem. alkaline phosphatase, which depends not only on secY but 262,1716-1719 also on secA and secD for export (46). 7. Kuhn, A., Kreil, G., and Wickner, W. (1987) EMBO J. 6, 501The transport function of our procoat hybrid was slightly 505 enhanced by reduction of the positive charge in thecleavage 8. Emr, S. D., and Silhavy, T. L. (1983) Proc. Natl. Aead. Sei. U. S. A. 80,4599-4603 region, either by replacing it with the procoat cleavage region 9. Kendall, D. A,, Bock, S. C., and Kaiser, E. T. (1986) Nature 321, or by changing thelysine at -2 to a neutral residue. However, 706-708 the most striking improvement occurred when a negatively 10. Briggs, M. S., and Gierasch, L. M. (1986)Adu. Protein Chem.38, charged portion of the mature coat proteinwas added next to 109-180 the procoat core and cleavage regions. Previous studies have 11. Bankaitis, V. A,, Ryan, J. P., Rasmussen, B. A. and Bassford, P. J., Jr. (1985) Cum. Top. Membr. Tramp. 24, 105-150 also found that a negative charge balance in the amino terminus of thematureproteinis a criticaldeterminant of 12. Kendall. D. A.. and Kaiser, E. T. (1988) J . Biol. Chem. 2 6 3 , 7261-7265 transport function;deletion of negative charges or addition of 13. Inouye, H., Michaelis, S., Wright, A., and Beckwith, J. (1981) J. positive chargesin the amino terminus of transported proteins Bacteriol. 146, 668-675 has been found to impair processing (34, 47, 48) and export 14. Ghosh, S. S., Bock, S. C., Rokita, S. E., and Kaiser, E. T. (1986) (34). Our findings demonstrate that an increase in the amino- Science 2 3 1 , 145-148 terminal negative charge of the mature protein canovercome 15. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloning:A Laboratory Manual, Cold Spring HarborLaboratory, the transport defects incurredby mismatches between signal Cold Spring Harbor, NY a result 16. Neidhardt, F. C., Bloch, P. L., and Smith, D. F.(1974)J.Bacteriol. peptide components anda given mature protein. Such reinforces the notion thatsignal peptides and mature proteins 119,736-747 cooperate inthetransport process, andthe efficiency of 17. van Wezenbeek, P. M. G. F., Hulsebos, T. J. M., and Schoenmakers, J. G. G. (1980) Gene (Amst.) 11, 129-148 transport is determined by a complex balance of their physical 18. Bedouelle, H., Bassford, P. J., Jr., Fowler, A. V., Zabin, I., and structural properties. Beckwith, J., and Hofnung, M.(1980) Nature 2 8 5 , 78-81 Although the exact means bywhich amino-terminal nega- 19. Movva. N. R.. Nakamura.. K... and Inouve, - . M. (1980) J. Mol. Biol. tive charges expedite transport is not understood, in the case 143,'317-328 of procoat they are thought to facilitate movement of the 20. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Aead. Sci. U. S. A. 74,5463-5467 protein through the bilayer in response to the electrical potential across the inner membrane(29). Because coat protein 21. Ito, K., Bassford, P. J., Jr., and Beckwith, J. (1981) Cell 24,707717 requires only signal peptidase for membrane insertion and 22. Russel, M., and Model, P. (1982) Cell 2 8 , 177-184 processing, this electrophoretic transport mechanism is pre- 23. Laemmli, U. K. (1970) Nature 227,680-685 sumed to be independent of any export machinery (29). The 24. Edman, P., and Begg, G. (1967) Eur. J. Biochen. 1,80-91 improved transport competence of our procoat-alkaline phos- 25. Matsudaira, P. (1987) J. B i d . Chem. 262, 10035-10038 '

Signal Peptide Hydrophobic Cores Are 26. Tommassen, J., van Tol, H., and Lugtenberg, B. (1983) EMBO J. 2, 1275-1279 27. Sjostrom, M., Wold, S., Wieslander, A., and Rilfors, L. (1987) EMBO J. 6,823-831 28. Bankaitis. V. A,. and Bassford. P. J.., Jr. (19841 J. Biol. Chem. . 259, 12193-12200 29. Wickner. W. (1988) Biochemistrv 27. 1081-1086 30. Chang, C. N., Inouye, H., Modei, P.,'and Beckwith, J. (1980) J. Bacteriol. 142, 726-728 31. Nakata, A., Shinagawa, H., and Rothman, F. G. (1987) in Phosphate Metabolism and Cellular Regulation i n Microorganisms (Torriani-Gorini, A., Rothman, F.G., Silver, S., Wright, A,, and Yagil, E., eds) pp. 139-141, American Society for Microbiology, Wash., D. C. 32. Nakata, A., Shinagawa, H., and Kawamata, J. (1979) FEES Lett. 105, 147-150 33. Perlman, D., and Halvorson, H. 0.(1983) J . Mol. Biol. 167,391409 34. Li, P., Beckwith, J., and Inouye, H. (1988) Proc. Natl. Acad. Sei. U. S. A. 85, 7685-7689 35. von Heijne, G. (1986) Nucleic Acids Res. 14, 4683-4690 36. Duffaud, G . , and Inouye, M. (1988) J. Biol. Chem. 263, 1022410228 37. Chou, P. Y., and Fasman, G. D. (1978) Annu. Reu. Biochem. 47, 251-276

Not Always Interchangeable

14485

38. Dierstein, R., and Wickner, W. (1985)J . Biol. Chem. 260,1591915924 39. Kyte, J., and Doolittle, R. F. (1982) J. Mol. Biol. 157, 105-132 40. McKnight, C. J.,Stradley, S. J.,and Gierasch, L. M. (1989) Abstract, Protein Folding Seminar, AAAS Annual Meeting,San Francisco, CA 41. Wallace, B. A., Cascio, M., and Mielke, D. L. (1986) Proc. Natl. Acad. Sci. U. S. A . 83, 9423-9427 42. Randall, L. L., and Hardy, S. J. S. (1986) Cell 46, 921-928 43. Park, S., Liu, G., Topping, T. B., Cover, W. H., and Randall, L. L. (1988) Science 239, 1033-1035 44. Crooke, E., Guthrie, B., Lecker, S., Lill, R., and Wickner, W. (1988) Cell 54, 1003-1011 45. Lill, R., Crooke, E., Guthrie, B., and Wickner, W. (1988) Cell 54, 1013-1018 46. Randall, L. L., Hardy, S. 3. S., and Thom, J. R. (1987) Annu. Rev. Microbiol. 41, 507-541 47. Boyd, D., Guan, C.-D., Willard, S., Wright, W., Strauch, K., and Beckwith, J. (1987) in Phosphate Metabolism and Cellular Regulation i n Microorganisms (Torriani-Gorini, A., Rothman, F. G., Silver, S., Wright, A., and Yagil, E., eds) pp. 89-93, American Society for Microbiology, Wash., D. C. 48. Russel, M., and Model, P. (1981) Proc. Natl. Acad. Sci. U. S. A . 78, 1717-1721