rprotein3

Protein Engineering vol.15 no.4 pp.337–345, 2002

Engineering a novel secretion signal for cross-host recombinantprotein expression

Nguan Soon Tan1,2, Bow Ho3 and Jeak Ling Ding1,4

1Department of Biological Sciences and 3Department of Microbiology,National University of Singapore, Singapore 1175434To whom correspondence should be addressed at: Department of BiologicalSciences, National University of Singapore, 10 Kent Ridge Crescent,Science Drive 4, Singapore 117543.E-mail: [email protected] address: Institut de Biologie Animale, Batiment de Biologie,Universite de Lausanne, CH-1015, Lausanne, Switzerland

Protein secretion is conferred by a hydrophobic secretionsignal usually located at the N-terminal of the polypeptide.We report here, the identification of a novel secretion signal(SS) that is capable of directing the secretion of recombinantproteins from both prokaryotes and eukaryotes. Secretionof fusion reporter proteins was demonstrated in Escherichiacoli, Saccharomyces cerevisiae and six different eukaryoticcells. Estrogen-inducibility and secretion of fusion reporterprotein was demonstrated in six common eukaryotic celllines. The rate of protein secretion is rapid and its expres-sion profile closely reflects its intracellular concentrationof mRNA. In bacteria and yeast, protein secretion directedby SS is dependent on the growth culture condition andrate of induction. This secretion signal allows a flexiblestrategy for the production and secretion of recombinantproteins in numerous hosts, and to conveniently and rapidlystudy protein expression.Keywords: broad hosts/protein expression/secretion signal

IntroductionWith innovative genomics technology, genes are being disco-vered faster than their functions can be characterized. As weenter the era of proteomics, the ability to rapidly produce largenumbers of proteins in a parallel manner becomes increasinglyimportant. Determining their functions requires numerousbiophysical (e.g. crystallization, NMR, MS) and functionalstudies (e.g. protein–protein interactions), each of which usesa different expression vector. Hindrances to these analysesinclude the arduous task of subcloning, problems with readingframe and Kozak sequence, as well as the downstreampurification protocols. A versatile system for transferring DNAfragments between vectors using the Cre-lox recombinasetechnology has been recently developed (Liu et al., 1998). Inaddition, the Sindbis expression system enables the rapid,high-level expression of heterologous proteins in a variety ofeukaryotic cell lines derived from mammalian, avian and insecthosts (Xiong et al., 1989). Recombinant proteins synthesizedin heterologous hosts may accumulate in one of three ‘compart-ments’: the cytoplasm, the periplasm or the extracellularmedium. Many overexpressed proteins from various originshave been purified from each of these locations. Wheneverpossible, secretion is the preferred strategy since it permitseasy and efficient purification from the extracellular medium.However, to date, each expression system needs specific

© Oxford University Press 337

tailoring to meet the stringent requirements for each proteinproduct to ensure correct folding, activity and desired yield.Furthermore, the flexibility of a common secretion signalsequence with which to secrete a wide variety of heterologousfusion proteins from various hosts into the extracellular mediumis not available.

Protein secretion is one of the most important issues ofprotein expression in fundamental processes of living cells.Successful protein secretion requires effective translocation ofthe protein across the endoplasmic recticulum or plasmamembrane. Proteins destined for secretion are targeted to themembrane via their respective secretion signals that are usuallylocated at the N-terminal of nascent polypeptides. These signalsdisplay very little primary sequence conservation. However,they all possess three general domains: an N-terminal regionthat varies widely in length, but typically, contain amino acidswhich contribute a net positive charge to this domain; a centralhydrophobic region made up of a block of seven to 16hydrophobic amino acids; and a C-terminal region that includesthe signal cleavage site (Nothwehr and Gordon, 1990; Pinesand Inouye, 1999). Since the principles of protein transloca-tion mechanism are evolutionarily conserved (Schatz andDobberstein, 1996), it is conceivable that there exists a secretionsignal that is operational in both prokaryote and eukaryote,viz, cross-host.

Our previous efforts to express and secrete the limulusFactor C, a highly complex serine protease, in Escherichiacoli, Pichia pastoris and COS cells using its native hydrophobicsignal, Saccharomyces cerevisiae α-mating factor orKluyveromyces lactis killer toxin secretion signal were unsuc-cessful (Roopashree et al., 1995, 1996; and unpublished data).Surprisingly, the secretion of the similar construct was achievedby a novel 15-residue hydrophobic secretion signal (SS) inDrosophila S2 cells (Tan et al., 2000a). Furthermore, varyingthe genes in the fusion or the tags, did not affect the high-level secretion and cleavage at the correct site (Tan et al.,2000a,b). Despite the origin of this signal, �80% of therecombinant proteins expressed by the heterologous insect hostwere localized in the extracellular medium. In this study,we demonstrate the versatility and functionality of SS forrecombinant protein expression and secretion in cross-hosts[E.coli, S.cerevisiae, higher eukaryotic cells—African greenmonkey kidney cells (COS-1), Chinese hamster ovary cells(CHO-B), fibroblast cells derived from Swiss mouse embryo(NIH/3T3), human cervical adenocarcinoma cells (HeLa), carpepithelial cells (EPC) and chinook salmon embryonic cells(CHSE)]. The expression and secretion of the recombinantproteins were performed using either a constitutive or aninducible promoter. In addition, we compared the secretionrate of reporter protein directed by SS and human secretedalkaline phosphatase (SEAP) signal, and assessed the efficiencyof secretion in different yeast media. This paper illustrates theengineering of SS to aid the production of secreted recombinantprotein for easier analysis. To the best of our knowledge, thisstudy documents the only known cross-host secretion signal.

N.S.Tan, B.Ho and J.L.Ding

338

Secretion signal for protein expression

Materials and methodsConstruction of secretory CAT and β-galactosidase expres-sion vectorsThe isolation and initial cloning of SS into pEGFP-N1, toyield pSSEGFP was described in Tan et al. (Tan et al.,2000a). Detailed sequences and cloning strategies of secretorychloramphenicol acetyltransferase (SSCAT) and β-galactosid-ase (SS-Gal) can be obtained from the corresponding author.The vector maps of various constructs are illustrated in Figure 1.

Cell culture and transfectionsCOS-1, NIH/3T3, HeLa and CHO-B cells were maintained inDMEM while EPC and CHSE were cultured in MEM. Allmedia were phenol-red free and supplemented with 10%charcoal/dextran-treated fetal bovine serum. Cells were trans-fected with 1 µg of SS-fusion construct:control vector in aratio of 8:2, by lipofectamine (Gibco BRL) as described bythe manufacturer.

For estrogen-induction experiment, cells were co-transfectedwith ERU-psp-SSCAT, pSGcER (chicken estrogen receptor)and pSEAP-Control as described in Tan et al. (Tan et al.,1999). For studies on the rate of secretion, better comparisonbetween the CAT and β-galactosidase ELISA were achievedby adapting to fluorescence assays using the DIG FluorescenceDetection ELISA (Boehringer Mannheim). SEAP was deter-mined fluorimetrically (LS-50B, Perkin Elmer) at Ex360nmand Em449nm. SSCAT and SS-Gal were detected at Ex440nmand Em550nm.

Expression of SSCAT in S.cerevisiaeThe construct pYEX-SSCAT was transformed into S.cerevisiaeDY150 (Chen et al., 1992). The transformants were selectedon synthetic minimal medium (MM) agar containing all therequired supplements except uracil. For expression analysis, a100 ml YEPD medium (2% yeast extract, 1% mycopeptone,2% D-glucose, 5� HTA: 100 mg each of histidine, tryptophanand adenine, pH 5.0) was inoculated with a single clone andgrown for 16 h at 30°C. Subsequently, 50 ml of the yeastcultures were grown independently for 72 h in 2�1 l baffledflasks containing either 200 ml of YEPD or MM. At indicatedtime intervals, 2 ml aliquots of culture were centrifuged toobtain yeast pellet and culture supernatant. The yeast cellswere lysed with glass beads (Roopashree et al., 1996), whereasthe culture medium was collected and frozen without any pre-treatment. The pH of the culture was adjusted to pH 5.0using 1 M potassium phosphate buffer (pH 8.0). SSCAT wasmeasured as described above.

Arabinose-induced expression of modified SS-β-lactamase inbacteriaTransformants of E.coli LM194 with pBADSSblactKana wereselected for by plating on LB agar containing 50 µg/mlkanamycin. For the ampicillin plate assay, LM194 competentbacteria were transformed and plated on ampicillin LB agar

Fig. 1. (a) The diagrammatic map of the pSSCAT vector. The expression of the SSCAT gene is driven by a strong constitutive promoter, CMV. The startATG codon of the CAT gene was mutated to CTG to ensure efficient translation initiation at SS. (b) pSS-Gal construct map. pSS-Gal used the backbone fromthe β-Gal-promoter (Clontech) except that SS was subcloned in-frame upstream of the β-galactosidase gene. (c) psp-SSCAT map. The psp-SSCAT harbors thesecreted SSCAT. The multiple cloning site (MCS) is as illustrated. (d) Map of the ERU-psp-SSCAT construct. The ERU-psp-SSCAT is similar to thepsp-SSCAT except that the 565 bp ERU of Xenopus vitellogenin B1 gene is subcloned upstream of the SV40 promoter. (e) pYEX-SSCAT vector map.The pYEX-SSCAT is the yeast vector expressing SSCAT. The vector backbone is pYEX-S1. The original K.lactis signal peptide was replaced by SS.(f) pBADSSblactKana vector map. The mutant β-lactamase gene, whose secretion is directed by SS is subcloned into the vector backbone of pBAD vector(Invitrogen). The SS-β-lactamase insert is regulated by the araBAD promoter. Another selective antibiotic resistance gene (kanamycin from pGFP-N3) wasused to replace the parental ampicillin resistance gene of pBAD vector.

339

plates containing 0.2% glucose or at increasing dosage ofarabinose. The expression and secretion of functional SS-β-lactamase was visualized as colony formation.

For liquid assay, 5 ml of RM medium (1� M9 salts, 2%casamino acids, 0.2% D-glucose, 1 mM MgCl2, 50 µg/mlkanamycin) was inoculated with either a single recombinantor untransformed LM194 colony. The induction procedurewas as described by the manufacturer (Invitrogen). Prior toinduction, a 1 ml aliquot of culture was removed, processedand designated the zero time point. The medium was clarifiedoff bacteria by centrifugation and sterile filtered using a0.22 µm membrane. The periplasmic space fraction wasisolated from the cell lysate (Laforet and Kendall, 1991). Boththe medium and periplasmic fraction were assayed forβ-lactamase activity (Cohenford et al., 1988).

Results

SS directs the secretion of reporter protein into culture mediumTo investigate whether SS can direct the secretion of commonreporter proteins from various eukaryotic hosts, fusion con-structs of SS with CAT and β-galactosidase driven byconstitutive CMV or SV40 promoter, were transfected into avariety of cell lines, namely COS-1, NIH/3T3, CHO-B, EPC,HeLa and CHSE. As illustrated in Figure 2, SSCAT, ssEGFPand SS-Gal were effectively secreted and accumulated in theculture medium of all the cell lines tested, although the amountof SS-fusion protein produced varied. Despite being dilutedin the culture medium, the secreted recombinant proteins weredetectable within 24 h, indicating high level expression.

Rapid secretion rate of SSCAT and SS-Gal compared toSEAPThe amount of SS-fusion protein secreted at various timeintervals was determined using a standard curve generatedfrom the positive control provided by the kits. The rate ofsecretion was determined by the gradient of the best-fit linewhen the amount of secreted protein was plotted against time.The mean rates of secretion of SSCAT and SS-Gal were15.8 fg/ml SSCAT/ng β-galactosidase/min � 0.12 fg/ml/minand 12.1 fg/ml SS-Gal/ng SEAP/min � 0.09 fg/ml/min,respectively (Figure 3). In comparison, SEAP was secretedat a rate of 4.76 fg/ml SEAP/ng β-galactosidase/min �0.06 fg/ml/min. The rate of SSCAT and SS-Gal secretion wasalmost 3-fold higher than SEAP. Thus, this indicates that thereis a rapid post-translational processing of the SS.

Inducible expression of SSCAT protein correlated with itsmRNA levelThe expression and secretion of SS-fusion proteins, in particularSSCAT, were also examined using an inducible promoter. Theestrogen-induced expression and secretion of SSCAT wasobserved in all the cell lines tested, although the amount ofSSCAT produced varied. Uninduced COS-1 cells exhibited


Fig. 2. (a) Western blot analysis of SSEGFP expression in COS-1 cells. Themajority of SSEGFP was secreted into the culture medium. This shows thatSS can direct secretion of a reporter gene, EGFP. For each sample, 30 µg oftotal protein from culture medium was used for electrophoresis. Lanes: M,molecular weight marker; 1, untransfected COS-1 cell culture medium,24 h; 2, culture medium, 24 h post-transfection; 3, culture medium, 48 hpost-transfection (b) Western blot analysis of SS-Gal using mouse anti-β-galactosidase. Fifty micrograms of culture medium was loaded andelectrophoresed in a 10% SDS–PAGE. Lanes: 1, molecular weight marker;2, day 5 medium; 3, day 4 medium; 4, day 3 medium; 5, day 2 medium;6, control medium; 7, 20 µg of cell lysate from day 5 culture. The westernblot was developed using goat anti-mouse-HRP and chemiluminescentsubstrate. (c) Secreted SSCAT expression was observed in all the six celllines tested (COS-1, NIH/3T3, CHO-B, EPC, HeLa, CHSE). SSCAT wasmeasured using ELISA. Values represent the mean of four independentexperiments.

only marginal increase in SSCAT over a period of 24 h. Forinduced COS-1 cells, the increase in SSCAT can be detectedas early as 2 h, reaching a peak of 7-fold increase at 12 hpost-induction (Figure 4a). Estrogen-induced expression ofSSCAT can also be observed in other vertebrate cell lines,namely NIH/3T3, CHO-B and EPC cells (Figure 4b).

To verify that the level of secreted SSCAT in the culture

340

Fig. 3. Rate of secretion of SSCAT and SS-Gal in comparison with SEAP.COS-1 cells were transfected with SS-fusion construct:control vectors. After36 h post-transfection, at intervals of 15 min over a period of 2 h, themedium from cells of each time point was removed and replaced with 1 mlof fresh medium. After the last time point, which should represent 0 min, anadditional 1 h incubation was employed for all cultures to avoid lowreading variations. At the end of incubation, the medium was clarified viacentrifugation. The rate of secretion was determined by the gradient of best-fit line when the amount of secreted protein was plotted against time. Thevalues for SSCAT and SEAP secreted were normalized by β-galactosidaseproduction. Similarly, the values for secreted SS-Gal were normalized withSEAP.

medium is directly proportional to changes in intracellularconcentration of SSCAT mRNA, a northern kinetic analysiswas performed under estrogen-stimulation. Figure 4c indicatesthat the level of SSCAT protein secreted into the culturemedium was directly proportional to changes in intracellularconcentration of SSCAT mRNA. The results indicate that thepreviously observed rapid secretion of SS-fusion proteinsdriven by constitutive promoters is not due to the strength ofthe promoter, but rather, the properties of SS as a secretionsignal. Thus far, we have demonstrated that SS is functionalin several common higher eukaryotic hosts, both mammalianand non-mammalian. In addition, under similar experimentalconditions, a higher level of SS-fusion proteins was detectedin the extracellular medium as compared to SEAP.

The novel SS can direct secretion of recombinant proteins inyeastAlthough, recombinant protein expression in yeast has itslimitations, it is still a favorable choice because it can becultivated readily in large-scale fermentation, with an advant-age of releasing relatively little extraneous protein material intothe medium and post-translational modifications of proteins. Tofurther examine the versatility of SS, the secretion of SS-fusion protein, SSCAT, driven by constitutive PGK promoterwas studied in two independent S.cerevsiae transformantscultured in two different media.

The SSCAT expression profile was monitored over 72 h inyeast grown in YEPD (rich medium) and MM (minimal


Fig. 4. (a) Inducible expression and secretion of the recombinant CAT reporter. COS-1 cells were cotransfected with ERU-psp-SSCAT, pSGcER and pSEAP-Control. Estrogen-induced expression of SSCAT was monitored over a period of 24 h upon addition of 10–7 M of E2. (b) Estrogen-inducibility observed inother eukaryotic cells. SSCAT was produced and secreted into the culture medium by NIH/3T3, CHO-B and EPC. Values are means of four independentexperiments. (c) Northern blot analysis of E2-induced SSCAT expression for ERU-psp-SSCAT. The levels of SSCAT secreted into the culture medium aredirectly proportional to changes in intracellular concentration of SSCAT mRNA. Actin was used to normalize the result.

medium). After 24–48 h of culture, the yeast transformantsgrown in MM secreted significantly less SSCAT in the medium.It is unlikely that the overall SSCAT expression was reducedin MM-cultured yeast since comparable SSCAT protein wasobserved in the yeast lysate of both MM and YEPD cultures.Interestingly, the decrease corresponds to a drop in the pH ofMM. Adjusting the pH back to 5, resulted in a tremendous

341

increase in secreted SSCAT. This effect is less pronounced inthe rich YEPD medium, probably because it supports high-density growth and has higher buffering capacity. The amountof SSCAT detected in both types of culture media wascomparable at 72 h (Figure 5).

It is noteworthy that although the amount of SSCAT inthe medium is ~50% that of yeast lysate, this is likely an


under-representation of the secreted SSCAT. The amount ofperiplasmic SSCAT was not determined, but was insteadincluded in the values of SSCAT in the yeast lysate.

The growth and expression profiles of SS-β-lactamase inbacteriaAmpicillin which belongs to the β-lactam group of antibiotics,binds to and inhibits a number of enzymes in the bacterialmembrane that are involved in the synthesis of the cell wall(Waxman and Strominger, 1983). The ampicillin resistancegene codes for β-lactamase, and is secreted into the periplasmicspace of the bacterium, where it catalyzes hydrolysis of theβ-lactam ring, with concomitant detoxification of the drug(Sykes and Mathew, 1976). As such, this imposes an absoluterequirement on the bacteria for both high level and rapidexpression/secretion of functional β-lactamase to ensure itssurvival. We next sought to investigate if SS can fulfill theserequirements necessary for the growth of the bacterial host.To this end, we have constructed a mutant β-lactamase, SS-β-lactamase, where its native secretion signal was replaced bySS in a construct, pBADSSblactKana. The expression of

Fig. 5. Expression profile of SSCAT in two different yeast transformants.Secretion of SSCAT into culture medium is significantly higher in the richYEPD medium. It is important to note that the cell lysate, in this instance,refers to SSCAT in both the cytosol and periplasmic space. Consequently,secretion of SSCAT is more efficient than that reflected by SSCAT detectedin the medium only.

Fig. 6. (a) Plate assay for SS-β-lactamase. A functional kanamycin genewas demonstrated by the ability of the bacteria to grow on kanamycin-containing LB agar. No bacterial colonies were observed for either 0.2%glucose or 0.0002% arabinose. As the concentration of arabinose inducerwas increased, smaller bacterial colonies were observed. (b) SS-β-lactamaseexpression profile in culture medium. Transformants induced with 0.0002%arabinose displayed the highest level of SS-β-lactamase in the medium.(c) SS-β-lactamase accumulation in the periplasmic space. Accumulation ofSS-β-lactamase in the periplasmic space displayed inducer dose-dependentexpression. Rapid and high accumulation of SS-β-lactamase in theperiplasmic space does not necessarily translate into higher amounts ofrecombinant protein in the culture medium.

342


Table I. Comparison of efficacy of SS with other secretion signals in four common expression hosts

Secretion signals Bacteriaa Yeastb Insectc Mammalian References

SS � � � � Current work; Tan et al., 2000a,b; Wang et al., 2001Growth hormone � � � Gray et al., 1985; Asakura et al., 1995Serum albumin � � � Coloma et al., 1992; Kirkpatrick et al., 1995Human placental alkaline phosphatase � � Golden et al., 1998Staphylococcal protein A � � Uhlen and Abrahmsen, 1989; Allet et al., 1997Honeybee melittin � Tessier et al., 1991Ecdysteroid UDP-glucosyltransferase � Laukkanen et al., 1996Tissue plasminogen activator � Farrell et al., 1999α-Mating factor � Brake et al., 1984; Kjeldsen, 2000PHO1 � Laroche et al., 1994K.lactis killer toxin � Baldari et al., 1987OmpA/T � Pines and Inouye, 1999Haemolysin � Blight and Holland, 1994; Chervaux et al., 1995Bacteriophage fd gene III � Rapoza and Webster, 1993

�, secretion competency of recombinant proteins from the host.aIncludes both Gram-positive and -negative bacteria.bIncludes S.cerevisiae, Schizosaccharomyces pombe and P.pastoris.cIncludes lepidoteran (i.e. baculovirus system) and Drosophila.

the SS-β-lactamase came under the control of an induciblearabinose-responsive promoter. As illustrated in Figure 6a, nocolonies were seen in the absence or presence of 0.0002%arabinose or 0.2% glucose alone, whereas dose-dependentarabinose-induced (0.002–0.2%) expression and secretion ofSS-β-lactamase permitted the bacterial transformants to surviveon ampicillin LB agar plates. Interestingly, the colony sizeappeared distinctively smaller with increasing levels of ara-binose.

To further examine the efficacy of SS directed β-lactamasesecretion, we decided to measure SS-β-lactamase activities viaa liquid assay. The pBADSSblactKana clone grown in RMmedium with 0.2% glucose (i.e. no induction) exhibited asimilar growth profile as the control LM194 host bacteria (datanot shown). Concomitant with the plate assay, no SS-β-lactamase activity can be detected in the culture medium andperiplasmic space (Figure 6b and c) of uninduced trans-formants. The addition of arabinose resulted in expression andaccumulation of SS-β-lactamase in the culture medium andperiplasmic space. Even more surprising is that the highestlevel of enzyme was detected when 0.0002% arabinose wasused (Figure 6c). This apparent conflict was due to the growth-inhibitory effect on the bacteria when induced at a highconcentration of arabinose (data not shown). The dose-depend-ent expression profile of SS-β-lactamase, however, was notobserved after 4 h. Similar results were obtained using theTOP10 strain of E.coli, although the overall protein expressionlevel decreased by ~20% (data not shown).

DiscussionThe fundamental basis for the search of a cross-host secretionsignal really lies in the efficacy between heterologous versushomologous secretion signals. Heterologous secretion signalsrefer to the use of this signal for the secretion of heterologousgene products, as well as from a different host from whichthe signal sequence was derived. In contrast, homologoussecretion signals refer to the secretion of its natural geneproduct from the same host species. Certainly, a cross-hostsecretion signal will have to satisfy three other importantcriteria: (i) this signal must confer secretion to gene productsof different origins (prokaryotic or eukaryotic); (ii) the func-tionality of this signal must extend beyond its original host;

343

(iii) the expression and secretion of the gene products mustbe of appreciable quantity and functional. Currently, no singlesecretion signal has been demonstrated to be effective inboth prokaryotic and eukaryotic host expression systems. Thecurrently available secretion signals have exhibited limitedfunctionality and/or non-compatibility for cross-host recombin-ant protein expression (Table I). Therefore, availability of acommon broad-host secretion signal is highly desirable. Themajor objective of this study was to evaluate the efficiency ofSS in directing cross-host expression and secretion of foreignproteins. Consequently, SS was subcloned upstream of threereporter protein genes—EGFP, CAT and β-galactosidase. Thesethree proteins were chosen because of their different size andorigin (prokaryotic versus eukaryotic).

Based on strict definition, no functional heterologous secre-tion signal was reported for bacterial use. Although theexpression and secretion of numerous heterologous genes,such as human superoxide dismutase (Takahara et al., 1988),have been successful in bacteria, most if not all bacterialexpression utilized secretion signals of prokaryotic origin(Table I). Perhaps, the closest example was that of humangrowth hormone (hGH). Gray et al. (Gray et al., 1985)compared the efficiency of export of hGH directed by eitherits own signal sequence or the E.coli Pho A signal sequence.Results indicated that the secretions are comparable with 72%of the hGH localized in the periplasm. However, the efficacyof the hGH signal in directing the secretion of heterologousproteins in bacteria has not been reported. In comparison, thepotential of SS to direct secretion of proteins in E.coli wasevaluated by the secretion of a modified SS-β-lactamase. Viaplate and liquid assays, we showed that the secretion is rapidand at least 50% of the protein is detectable in the extracellularmedium upon induction. However, high doses of the inducer,arabinose, led to lower secreted product. Overloading theexport machinery may result from inefficient secretion of aforeign protein because the protein is expressed at levels thatsimply exceed cellular capacity. This is the first report of afunctional heterologous signal sequence in bacteria that permitsappreciable yield of secreted recombinant protein.

The first heterologous secretion signal for yeast was thehuman serum albumin (hHSA). This human secretion signalworks well in yeast, producing ~50% of the hHSA in yeast


fermentation media (Sleep et al., 1990). This signal results notonly in the hHSA secretion but also the secretion and desiredprocessing of other heterologous genes, such as humanimmunodeficiency virus (HIV) gp120 (Lasky et al., 1986) andsomatostatin (Itakura et al., 1977). Again, the functionalityof hHSA signal in bacteria was not reported. Interestingly,expression of hGH in yeast results in properly processed hGHin yeast media, suggesting that the signal recognition is notflawed. However, only 10% of the expressed protein is secretedwhereas 90% of hGH remains cell-associated and retains theentire signal sequence (Hitzeman et al., 1984). In comparison,we used SS to direct the secretion of a prokaryotic protein,SSCAT. As shown in Figure 5, at least 50% of the proteinwas secreted into the yeast media. Unlike, the hHSA signalsequence, SS is applicable in bacteria. It is worth noting thatin rapidly growing expression hosts, such as that of E.coli andS.cerevisiae, the rate of secretion is greatly influenced by theirgrowth conditions. Consequently, for optimal secretion ofrecombinant protein via SS, in rapidly growing expressionhosts, a compromise must be struck between growth conditionand concentration of the inducer, in order to regulate the rateof recombinant protein production and its secretion.

Many secreted eukaryotic proteins are efficiently processedin mammalian expression host via their native signal sequences.Hence, a more comprehensive study was done with SS. TheSS is able to direct secretion of both prokaryotic (SSCAT andSS-β-galactosidase) and eukaryotic proteins (EGFP), regardlessof protein size. Moreover, the rate of secretion of heterologousproteins is at least 3-fold faster than the SEAP native signalsequence. Taken together, SS is the only signal sequenceknown to date that is functional in all four common expressionhosts (Table I).

What makes SS such an efficient universal secretion signal?SS is capable of cross-host secretion for several reasons. First,it has the three domains typified in all secretion signals andthe presence of small amino acid residues at position –1 and –3 of the cleavage site (Jain et al., 1994). Secondly, the chargeto hydrophobicity ratio between the N-terminal domain andhydrophobic core, which is important for directing the proteinto the membrane (Rusch et al., 1994; Izard et al., 1996),represents a compromise of the requirements needed by boththe prokaryotic and eukaryotic hosts. Lastly, while manycurrently available secretion expression vectors also satisfythe first criteria, few possess the optimal ratio with respect tocriteria two, and none of them address the issue of sequencesbeyond the cleavage site, i.e. C-terminal, necessary for effectiveand homogenous cleavage and thus secretion. It is conceivablethat the C-terminal sequences are able to tolerate more degener-acy and hence the apparent redundancy to highlight thiscriteria. Effective cross-host secretion, in our case, requiresthis neglected criterion to be resolved. Initial attempts toreduce and/or remove the post-cleavage remnant residuesresulted in non-secreted recombinant protein (unpublisheddata). This compromise of post-cleavage six residues in therecombinant proteins is highly unlikely to alter the recombinantprotein functions. Comparatively, the post-cleavage remnantresidues of SS are significantly smaller than GST or GFP andpossess lesser charge than the 6 histidine tag. While this workclearly documents SS as a cross-host secretion signal, thefunctionality of the secreted recombinant protein produced byany particular host will also strongly depend on other factorssuch as post-translational modifications and intrinsic propertiesof the protein to be expressed.

344

This study reports the identification and development of across-host secretion signal. Its ability to direct recombinantprotein secretion was evaluated with SS-fusion reporter proteinsin various hosts—higher eukaryote, yeast and bacteria. Weenvision that fusion of the SS to recombinant genes will proveto be a valuable tool for efficient protein secretion in a broadheterologous host expression system. This secretion signal canbe incorporated into the ‘donor vector’ of various multi-vectorcloning systems, such as Gateway™ (Gibco BRL) and Echo™Cloning (Invitrogen), which can then be transferred intovarious host expression vectors for expression and secretionof recombinant proteins. This secretion signal can also beincorporated into various reporter assay systems for rapid, andminimal set-up reporter gene analyses. While an exhaustivescreen of all proteins is beyond the scope of this study, duringthe process of preparing this paper, the SS has been furtherevaluated by other researchers and was proven to yield variedsuccess in the secretion of other recombinant proteins, forexample, in Drosophila S2 cells, Sf9 cells (Wang et al., 2001)and E.coli (unpublished data).

AcknowledgementsWe thank Professor W.Wahli for Xenopus Vtg B1 ERU, and ProfessorP.Chambon for pSG cER. This work was funded by NUS GrantRP3999900/A and NSTB Grant LS/99/004.

ReferencesAllet,B., Bernard,A.R., Hochmann,A., Rohrbach,E., Graber,P., Magnenat,E.,

Mazzei,G.J. and Bernasconi,L. (1997) Protein Expr. Purif., 9, 61–68.Asakura,A., Minami,M. and Ota,Y. (1995) Biosci. Biotechnol. Biochem., 59,

1976–1978.Baldari,C., Murray,J.A., Ghiara,P., Cesareni,G. and Galeotti,C.L. (1987) EMBO

J., 6, 229–234.Blight,M.A. and Holland,I.B. (1994) Trends Biotechnol., 12,450–455.Brake,A.J., Merryweather,J.P., Coit,D.G., Heberlein,U.A., Masiarz,F.R.,

Mullenbach,G.T., Urdea,M.S., Valenzuela,P. and Barr,P.J. (1984) Proc. NatlAcad. Sci. USA, 81, 4642–4646.

Chen,D.C., Yang,B.C. and Kuo,T.T. (1992) Curr. Genet., 21, 83–84.Chervaux,C., Sauvonnet,N., Le Clainche,A., Kenny,B., Hung,A.L., Broome-

Smith,J.K. and Holland,I.B. (1995) Mol. Gen. Genet., 249, 237–245.Cohenford,M.A., Abraham,J. and Medeiros,A.A. (1988) Anal. Biochem., 168,

252–258.Coloma,M.J., Hastings,A., Wims,L.A. and Morrison,S.L. (1992) J. Immunol.

Methods, 152, 89–104.Farrell,P.J., Behie,L.A. and Iatrou,K. (1999) Biotechnol. Bioeng., 64, 426–433.Golden,A., Austen,D.A., van Schravendijk,M.R., Sullivan,B.J., Kawasaki,E.S.

and Osburne,M.S. (1998) Protein Expr. Purif., 14, 8–12.Gray,G.L., Baldridge,J.S., McKeown,K.S., Heyneker,H.L. and Chang,C.N.

(1985) Gene, 39, 247–254.Hitzeman,R.A., Chen,C.Y., Hagie,F.E., Lugovoy,J.M. and Singh,A. (1984) In

ed. Arthur P.Bollon Recombinant DNA Products: Insulin, Interferon andGrowth Hormone. CRC Press, Boca Raton, FL.

Itakura,K., Hirose,T., Crea,R., Riggs,A.D., Heyneker,H.L., Bolivar,F. andBoyer,H.W. (1977) Science, 198, 1056–1063.

Izard,J.W., Rusch,S.L. and Kendall,D.A. (1996) J. Biol. Chem., 271, 21579–21582.

Jain,R.G., Rusch,S.L. and Kendall,D.A. (1994) J. Biol. Chem., 269, 16305–16310.

Kirkpatrick,R.B., Ganguly,S., Angelichio,M., Griego,S., Shatzman,A.,Silverman,C. and Rosenberg,M. (1995) J. Biol. Chem., 270, 19800–19805.

Kjeldsen,T. (2000) Appl. Microbiol. Biotechnol., 54, 277–286.Laforet,G.A and Kendall,D.A. (1991) J. Biol. Chem., 266, 1326–1334.Laroche,Y., Storme,V., De Meutter,J., Messens,J. and Lauwereys,M. (1994)

Biotechnology, 12, 1119–1124.Lasky,L.A., Groopman,J.E., Fennie,C.W., Benz,P.M., Capon,D.J.,

Dowbenko,D.J., Nakamura,G.R., Nunes,W.M., Renz,M.E. and Berman,P.W.(1986) Science, 233, 290–212.

Laukkanen,M.L., Oker-Blom,C. and Keinanen,K. (1996) Biochem. Biophys.Res. Commun., 226, 755–761.

Liu,Q., Li,M.Z., Leibham,D., Cortez,D. and Elledge,S.J. (1998) Curr. Biol.,8, 1300–1309.


Nothwehr,S.F. and Gordon,J.I. (1990) BioEssays, 12, 479–484.Pines,O. and Inouye,M. (1999) Mol. Biotechnol., 12, 25–34.Rapoza,M.P. and Webster,R.E. (1993) J. Bacteriol., 175, 1856–1859.Roopashree,S.D., Chai,C., Ho,B. and Ding,J.L. (1995) Biochem. Mol. Biol.

Int., 35, 841–849.Roopashree,S.D., Ho,B. and Ding,J.L. (1996) Mol. Marine Biol. Biotechnol.,

5, 334–343.Rusch,S.L., Chen,H., Izard,J.W. and Kendall,D.A. (1994) J. Cell Biochem.,

55, 209–217.Schatz,G. and Dobberstein,B. (1996) Science, 271, 1519–1526.Sleep,D., Belfield,G.P. and Goodey,A.R. (1990) Biotechnology, 8, 42–46.Sykes,R.B. and Mathew,M. (1976) J. Antimicrob. Chemother., 2, 115–157.Takahara,M., Sagai,H., Inouye,S. and Inouye,M. (1988) Bio/Technology, 6,

195–198.Tan,N.S., Frecer,V., Lam,T.J. and Ding,J.L. (1999) Biochim. Biophys. Acta,

1452, 103–120.Tan,N.S., Ho,B. and Ding,J.L. (2000a) FASEB J., 14, 859–870.Tan,N.S., Ng,P.M.L., Yau,Y.H., Chong,P.K.W., Ho,B. and Ding,J.L. (2000b)

FASEB J., 14, 1801–1813.Tessier,D.C., Thomas,D.Y., Khouri,H.E., Laliberte,F. and Vernet,T. (1991)

Gene, 98, 177–183.Uhlen,M. and Abrahmsen,L. (1989) Biochem. Soc. Trans., 17, 340–341.Wang,J., Ho,B. and Ding,J.L. (2001) Biotechnol. Lett., 23, 71–76.Waxman,D.J. and Strominger,J.L. (1983) Annu. Rev. Biochem., 52, 825–869.Xiong,C., Levis,R., Shen,P., Schlesinger,S., Rice,C.M. and Huang,H.V. (1989)

Science, 243, 1188–1191.

Received May 25, 2001; revised December 18, 2001; accepted January 4, 2002

345

Documents

rprotein3