Proc. Natl. Acad. Sci. USA
`Vol. 77, No. 6, pp. 3369-3373, June 1980
`Biochemistry
`
`Eukaryotic signal sequence transports insulin antigen in
`Escherichia coli
`
`(rat preproinsulin/hybrid signal sequences/secretion/cloning vectors)
`
`KAREN TALMADGE, STEPHEN STAHL*, AND WALTER GILBERT
`
`Department of Biochemistry and Molecular Biology, Harvard University, Cambridge, Massachusetts 02138
`
`Contributed by Walter Gilbert, March 28, 1980
`
`ABSTRACT We made a series of plasmids with unique Pst
`restriction sites within or near the DNA that encodes the fini-
`cillinase si
`al sequence. Inserted DNA can be read in all
`ee
`frames bo within and immediately after the signal
`uence.
`We cloned Pst-terminated DNA copies of the structtiredl infor-
`mation for rat proinsulin and
`roinsulin into these plasmids,
`forming a large number of hybrri penicillinase (bacterial) and
`insulin éeukaryotic) signal sequences. We then com ared the
`levels 0 insulin antigen in the Escherichia coli
`rip asm with
`those inside the cells. We conclude that either t e bacterial or
`the eukaryotic signal is sufficient to transport rat insulin antigen .
`into the periplasmic space.
`
`Proteins must pass through a cell membrane in order to function
`outside the cell. With one known exception (1), all secreted
`proteins, both eukaryotic and prokaryotic, have an amino-
`terminal extension, or pre-sequence, that is removed as or after
`the protein cro$es the membrane (2). The amino acid sequences
`of more than 30 pre-sequence peptides (3) share one obvious
`structural feature: a stretch of 5-10 highly hydrophobic residues
`near the middle of the peptide. The signal hypothesis (2, 4)
`proposes that these hydrophobic amino-terrninal extensions
`serve to bind the protein to the membrane and then to lead the
`protein through.
`Direct evidence establishes that the bacterial pre-sequence,
`or signal sequence, has an essential role in protein transport.
`Emr et al. (5) and Bedoulle et al. (6) characterized in two cases
`a number of mutations that lead to the accumulation of the
`
`mutant protein in the cytoplasm as the pre-protein. In each case,
`the mutations result in amino acid replacements in the signal
`sequence, most of which are changes from a hydrophobic to a
`charged residue.
`Recombinant DNA technology attempts to produce higher
`cell proteins in bacteria. Such proteins are simpler to detect and
`purify if they are secreted from the cell. Villa-Komaroff et al.
`(7) cloned a CDNA copy of the rat proinsulin gene into the Pst
`restriction site of plasmid pBR322 (8). The Pst site encodes
`amino acids 181-182 of E. colt penicillinase (9), a periplasmic
`protein with a 23 amino acid signal sequence (9, 10). Thus,
`when the proinsulin sequence was inserted into the middle of
`the penicillinase gene, a fused protein was created which served
`to carry most of the insulin antigen to the periplasmic space
`(7)-
`'
`This successful transport of the penicillinase—insulin hybrid
`molecule prompted two overlapping experimental lines. First,
`we wanted to use the recombinant DNA techniques to alter
`pBR322 to create a set of sites for cloning closer to and within
`the signal sequence region and thus eliminate as much of the
`extraneous bacterial protein as possible. Second, we wanted to
`examine altered signal sequences created by such clonings to
`answer direct questions about the role of those sequences in
`protein secretion.
`
`The publication costs of this article were defrayed in part by page
`charge payment. This article must therefore be hereby marked “ad-
`vertisement" in accordance with 18 U. S. C. §l734 solely to indicate
`this fact.
`
`MATERIALS AND METHODS
`
`Bacterial Strains. E. coli K-12 strain HBl0l (hrs ‘, him‘,
`recA ‘, gal “, str', B1‘) was obtained from Herb Boyer; E. coli
`K-12 strain FMAIO (F‘, su‘, gal ‘, str’, thyA ‘, deo‘, hrs ‘,
`hrm+) was provided by Fred Ausubel and lysogenized by
`Stephanie Broome with X (21357; E. coli K-12 strain PRl3
`(pnp 13, mal9, thrl, leuB6, lac lambda 1, xyl7, mtl2, malAl,
`strAl32, lambda“, lambda‘) was from the Yale strain li-
`brary.
`Enzymes. Restriction enzymes and DNA polymerase I
`(Klenow fragment) were purchased from New England
`BioLabs Polynucleotide kinase was purchased from Boehringer
`Mannheim. BAL3l was the gift of Horace Gray. T4 DNA ligase
`was a gift of A. Poteete and Stewart Scherer.
`DNA. pBR322 was obtained originally from Herb Boyer;
`p119 and p147 were the gift of Peter Lomedico. Plasmid was
`prepared as described (12) and (13). Pst linker was purchased
`from Collaborative Research and treated with kinase under
`standard conditions (14).
`Ligations and Transformations. Ligations (15) and trans-
`formations (16) were done under standard conditions.
`Exonuclease. The ends of insulin insert 1947 were digested
`with the double-stranded exonuclease BAL3l in the following
`mixture (17): 2.5 ag of DNA, 2 units of BAL3l in 150 pl of 20
`mM Tris-HCI (pH 8), 12.5 mM MgSO4, 12.5 mM CaCl2, 0.2 M
`NaCl, and 1 mM EDTA; incubation was for 45 sec at 15°C.
`Radioimmunoassays. Two-site solid-phase radioimmuno-
`assays were performed as described (18) with minor modifi-
`cations (7). FMAl0/X CI357 was transformed with signal se-
`quence plasmids containing insulin gene inserts and induced
`at 42°C. Standard liquid radioimmunoassays were performed
`as described (19). Aliquots of cell fractions were preincubated
`for 1-3 hr with an amount of guinea pig anti—porcine insulin
`IgC fraction [prepared as described (l8)] sufficient to complex
`75% of the added labeled insulin.
`Cell Lysis and Fractionation. FMA10/X c1357 and PRl3
`bearing signal sequence plasmids expressing insulin antigen
`were grown in 100 ml of glucose minimal medium (supple-
`mented with 40 ug of thymine per ml for FMAl0) (14) at 34°C
`to OD55o = 0.24—0.4. Harvested cells were washed in 1 ml of
`100 mM Tris-HCI, pH 8/20% sucrose. Whole-cell lysis was
`performed exactly as described for plasmid preparation by the
`method of Clewell (12). Cell fractionation was by spheroplast
`formation and lysis (M. Russel, personal communication).
`Washed cells were resuspended in 900 pl of Tris/sucrose, in-
`cubated 30 min with 100 pl of lysozyme (5 mg/ml in 20 mM
`EDTA), and pelleted. The pelleted spheroplasts were washed
`gently with Tris/sucrose, resuspended in 100 [.l.l of Tris/sucrose
`(by using a glass rod), lysed by the addition of 850 pl of 0.3%
`Triton/150 mM Tris-HCI, pH 8/0.2 M EDTA (12), and cen-
`trifuged for 1 hr at 17,000 rpm in a Sorvall SA-600 rotor. A
`
`3369
`
`‘ Present address: Biogen S.A., 3, route de Troinex, 1227 Geneva,
`Switzerland.
`Mylan v. Genentech
`Mylan v. Genentech
`IPR2016-00710
`IPR2016-00710
`Genentech Exhibit 2092
`Genentech Exhibit 2092
`
`

`
`3370
`
`Biochemistry: Talmadge et al.
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`7|
`
`3
`
`pKTl7|
`pKTI90
`pKT2l8
`pKT22s
`OKT234
`pKT24I
`pKT279
`pxrzso
`pKT207
`
`’-°___~
`Ecol
`T I////In
`
`ooomu
`
`.3‘
`
`P Q0
`
`IND
`
`28
`Met:SerI1eG1nHisPheArgVa1A.l.aLé3I1el>roPhePheA1aA1aPheCysLeuP§'gVa1PheA1aHisProG1u'l'hrLeu. . .fig€Pr:oA1aA1aHet:.
`pBR322 ATGAGTA’l'l‘CAACAT'I'l‘CCGTG'I‘CGCCCPTATPCCCTTP1TTGCGGCAT1TTGCCTPCC1C1111TGC1CACCCAGAAACGCTG. . .A’l‘GCC'l‘GCAGCAA’l'G
`Pst:
`
`. .
`
`Metserr1eG1ilAZaA1aA1aMet. . .
`pKT218 ATGAGTATICAAGCTGCA GCAATG. . .
`Pst
`
`7
`MetserI1eG1nHisPheArgLeuGZnGZn. . .
`PKTZZ6 ATGAGTATTCAACATTTCCGGCTGCA GCAATG. . .
`PSC
`
`MetSerI 1eG1nHisPheArqva1A1gArgCysSer-Asn. . .
`pK'r234 ATGAGTAHCAACAMTCCGTGTCGCCCGCTGCAGCAATG. . .
`Pst:
`
`HetSerI1eG1nHisPheArgVa1A1aLeuI1ePi."gLeuGZnGZn. . .
`pKT24 1 ATGAGTAT'I‘CAACA'I"I'l‘CCG'1‘G'!’CGCCCTTA'1'l'CCGCTGCA GCAATG . . .
`Pst
`
`MetSerI1eG1nHisPheArgVa1A1aLeuI1eProPhePheA1aA1aPheCysLeuPxoVa1PheA1aH¥gArgCysSeI-Asn. . .
`pKT279 A’l‘GAGTATTCAACA’I"I"I‘CCG'I‘G'RZGCCC1TAT1CCCTHTPPGCGGCATPTPGCCTICCTGI'I'PTTGCTCACCGCTGCAGCAATG. . .
`Pst
`
`25
`.
`Metserl 1eG1nHisPheAtgVa1A1aLeuI1eProPhePheA1aA1aPheCysLeuProVa1PheAla!-l1sProLeuG ZnG Zn. . .
`pK'I‘280 ATGAGTATTCAACATTTCCGTGTCGCCCTTAT'I'CCCTPTI'l'PGCCGCATTl'PxCTTCCTGTT1'ITGCTCACCCGCTGCAGCAATG. . .
`Pst:
`
`MetserI1eG1nHisPheArgVa1A1aLeuI1eProPhePheA1aA1aPheCysLeuProva1PheA1aHisProG1u'IԤ;A ZaA1aA1aMet . . .
`pKT287 ATGAGTA'I'I‘CAACA'I‘TTCCGTGTCGCCCTTATTCCC'l'1"l'I"I'I‘GCGGCATPTPGCCTICCTGITHTGCTCACCCAGAAACGGCTGCAGCAATG. . .
`Pst
`
`FIG. 1. Deletion map of pBR322 penicillinase gene and sequence of derivative plasmid signal sequence regions (constructions to be described
`elsewhere). DNA regions that encode protein are represented as follows: penicillinase signal sequence, hatched; mature penicillinase protein,
`black. The derivatives were deleted from the Pst to the signal sequence coding region and the Pst site (C-T-G-C-A-G) was re-created by insertion
`of a Pst linker whose sequence is G-C-T-G-C-A-G-C. The bases donated by the linker on that strand are indicated in italics. The last wild-type
`penicillinase amino acid is indicated by the number of its wild-type position above it. The amino acids encoded by the inserted Pst linker are
`in italics. The arrows indicate the site of cleavage for maturation of wild-type prepenicillinase to penicillinase.
`
`membrane fraction was prepared from pelleted spheroplasts
`by two methods. Spheroplasts were resuspended in 1 ml of 10
`mM Tris-HCI, pH 8/5 mM MgCl2/5 mM dithiothreitol/0.2 M
`KCl (20) and sonicated three times for 10 see each on ice. Al-
`ternately, the spheroplasts were resuspended in 100 pl of
`Tris/sucrose and lysed with 900 pl of distilled water (M. Russel,
`personal communication). Both lysates were centrifuged for
`1 hr at 35,000 rpm in a Beckman SW 60 rotor. The membrane
`pellet was resuspended in Triton buffer with a Dounce ho-
`mogenizer.
`DNA Sequence Analysis. Plasmid prepared from 5 ml of
`cells was 3’-end labeled in the presence of 20 ;1M GTP and 2
`p.M [oz-32P]ATP at 15°C for 4 hr (21). Sequence analysis was
`by the method of Maxam and Gilbert (14).
`Recombinant DNA. All manipulations involving cells with
`insulin~gene plasmids were done under P2 containment ac-
`cording to the National Institutes of Health guidelines issued
`22 December 1978.
`
`RESULTS
`
`Fig. 1 shows a deletion map of nine derivatives of pBR322 and
`the sequences of seven of these plasmids that have useful Pst
`restriction sites in the penicillinase signal sequence coding re-
`gion (unpublished data). Four of these plasmids—pKT2l8,
`pKT226, pKT234, and pKT24l—code for 4, 7, 9, and 12 signal
`sequence amino acids, respectively. Three of the plasmids-
`pKT279, pKT280, and pKT287—-code for the entire signal
`sequence as well as for one, two, and four amino acids of the
`mature penicillinase respectively. Within each group, there are
`plasmids in all three reading frames.
`Fig. 2 shows the restriction maps and 5’-end sequences of the
`p119 and p147 cDNA Pst-ended gene fragments of rat
`preproinsulin and proinsulin isolated and sequenced by Villa-
`Komaroff et al. (7). We used BAL3l to chew back the ends of
`the pIl947 insert (derivation explained in figure legend), ligated
`the pieces to kinase-treated Pst linker, and cloned Pst-diges-
`ted pieces into the Pst site of pBR322 in HBl01. We used the
`
`

`
`Biochemistry: Talmadge et al.
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`3371
`
`Put
`
`-2:
`
`Ava
`
`(qua)
`Avon 5
`"
`
`+06 HIM PI!
`---|—> pug
`
`
`
`_21
`
`
`uGZnGZyG ZyGZyG ZyG Zy'l'rpHet.AtgPheI.euProLeuLeuA1aLeuLeuVa1Leu'rrpG1uPx'oLysProA1aG1nA1aPheVa1LysG1nHisIeuCys . . .
`CTGCAGGGGGGGGGGGGGGGG1GGATGCGC'l'K3CTGCCCCTGCTGGCCC'l'GI'lCG'lCC'1C'IGGGAGCCCAAGCC'I'GCCCAGGC'l"l"l"IC'lCAAACAGCACCT'I"I'GT. . . 19, 1947
`Pst
`-21
`A Za.A MA MC ZyTrpHetArgPheLeuProLeuLeuA1aIeu.LeuVa1Leu'1‘rpG1uProLysProA1aG1nA1aPheVa1LysGlnflisLeuCys . . .
`C'lCCAGCGGGGTGGA'l.‘GCGCHCCTGCCCC1GC'MGCCC1GCNGNCmmGGAGCCCMGCCTGCCCAGGm1T1CKMACAGCACCHmT. . . CB6
`
`Insert
`
`Pst
`
`_7
`LeuGZnArgG1uProLysProA1aG1nA1aPheVa1LysG1nHisI..euCys . . .
`C'1GCAGCY3GGAGCCCAAG$1GOCCAGGCTl'l'l'G1CAAACAGCACCT1'R3T. . . CB15
`Pst
`ArgCyaSerG1IrProLysProA1aG1nA1aPheVa1LysG1nHisI.euCys . . . CB16
`C'lGCAGCGAGCCCAAGCC'I’GCCCAGGCTTT'R31CAAACAGCACC1‘T'ICT. . .
`Pst
`
`Amzaazyazgazgozyazgczycfluisneucys. . .
`C1CCA CAGCACCTT'lCT_. . . 47
`Pst
`
`PM +4 Ava]!
`
`mu
`
`+09 "W P"
`
`pl47
`
`FIG. 2. Restriction map of rat preproinsulin (p119) and proinsulin (pI47) Pst inserts (7) [1947 is a recombinant between the 19-insert 5’
`end and the 47-insert 3’ end at the first Ava II site to remove a mutant glycine encoded in the 19 insert (22)]; sequences at the 5’ end of these
`inserts and the digested derivatives of 1947 insert. Bases in the digested 1947 insert sequences in italics have been donated by an inserted Pst
`linker. The first wild-type amino acid is indicated by the number of its wild-type position above it. Amino acids in italics were created by G-C-tailing
`during the original isolation of p119 and p147 (7) or by the insertion of a Pst linker. Arrows indicate the site of cleavage for maturation of
`preproinsulin to proinsulin.
`
`Maxam and Gilbert C+T and G+A reactions (14) to analyze
`inserts digested with Ava II and 3’-end labeled (21). With these
`Pst fragments, we can insert the gene for rat preproinsulin in
`two reading frames and the gene for rat proinsulin in all three
`reading frames in the penicillinase sequence.
`Table 1 lists the insulin constructions. We name the hybrid
`protein products with a lower case
`and a pair of numbers:
`the first refers to the last prepenicillinase wild-type amino acid
`before the amino acids encoded by the insertion of the Pst site,
`and the second refers to the first amino acid of the preproinsulin
`(negative numbers) or proinsulin (positive numbers) sequence.
`Table 2 shows the amino acid sequence of each hybrid protein
`from the first methionine to proinsulin amino acid 7. In each
`case, there is a minimum of three amino acids between the last
`amino acid of the penicillinase signal sequence and the first
`amino acid of the insulin portion; these extra amino acids are
`in italics in Figs. 1 and 2.
`Expression of Insulin Antigens. To explore the transport
`of the various fusions of proinsulin and preproinsulin to the
`modified signal sequences, we identified clones that were ex-
`pressing rat proinsulin antigen with a two-site solid-phase ra-
`100
`
`dioimmunoassay (18) and then fractionated the cells into a
`periplasmic and a cytoplasmic/membrane fraction. A lyse-
`zyme/EDTA treatment in hypertonic sucrose released the
`periplasmic proteins. After washing the resulting spheroplasts
`in Tris/sucrose, we lysed them with Triton (12) and removed
`the chromosomal DNA by centrifugation at 30,000 X g. We
`assayed the insulin antigen in these two fractions and compared
`those results to the values found in a Triton lysate of the whole
`cells. Less than 1% of the cells were lysed during the formation
`of the spheroplasts, as shown by [3-galactosidase assays (23) on
`cells induced with IPTG.
`
`We measured insulin antigen by a standard radioimmune
`assay (19). Fig. 3 shows typical a.ssays for each of three different
`constructions as well as a standard curve. Naturally, the bac-
`terial material is not identical to mature insulin, and the com-
`petition tails off earlier. To calculate the number of molecules
`per cell, we used the amount of cell extract required for 50%
`inhibition of the binding of radiolabeled antigen, the number
`of input cells, and the standard insulin curve. Table 1 collects
`all the results; Table 2 summarizes them. Some curves had to
`be extrapolated to the 50% point; such data, indicated by pa-
`
`
`
`80
`
`G’:O
`
`3S
`
`95inhibition
`
`1
`
`2
`
`4
`
`8
`
`16
`
`32
`
`64
`
`4
`2
`Cell extract. 541
`
`8
`
`16
`
`32
`
`4
`
`8
`
`16
`
`38
`
`0.2 0.4 0.8 1.6
`Insulin. n:
`
`Inhibition curves from radioimmunoassays for insulin antigen from cell extracts (Left) and a standard curve (Riglit). In the cell
`FIG. 3.
`extract assays, solid lines represent the contents of the periplasm and dashed lines represent the contents of the cytoplasmic/membrane frac-
`tion.
`
`

`
`3372
`
`Biochemistry: Talmadge et al.
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`Table 1.
`
`Insulin constructions
`
`Plasmid
`
`Insulin
`
`Host
`
`Molecules per cell
`P
`C/M WC % PS
`
`pI47
`
`p287.47
`
`p280.1947
`
`p241.1947
`
`p2l8.CB6
`
`i27/+4
`
`i181/+4 FMA10 >70‘
`PR13
`—
`PR13
`—
`6501
`—
`1424
`1517
`124
`132
`156
`160
`—
`1320
`1592
`—
`18
`—
`365
`
`i25/-21
`
`FMA10
`
`i12/-21
`
`FMA10
`
`PR13
`
`i4/-21
`
`FMA10
`
`PR13
`
`— ~100* >70
`—
`980 —
`— 5903 —
`(292)
`— >96
`— 1533 —
`(126)
`— >92
`(143)
`— >91
`(18)
`— >88
`(10)
`— ' >93
`(29)
`— >84
`(11)
`—— >93
`—
`1555 —
`(298)
`— >82
`105
`— 94
`——
`23 -
`(4)
`— >82
`——
`368 —
`(37)
`>91
`
`hybrid proteins produced by two fusions that inserted a charged
`amino acid after a complete penicillinase signal (i25/-7 and
`i24/-7) appeared half in the periplasm and half in the cyto-
`plasmic/membrane fraction.
`In an attempt to determine whether or not those proteins that
`were not fully transported were in the membrane, we sonicated
`the spheroplast pellets (20) and collected the membranes by
`centrifugation at 100,000 X g. Only about 10% of the total in-
`sulin antigen in these cells was associated with the membrane
`pellet (resuspended in Triton). In addition, we gently lysed the
`spheroplast pellet from cells producing i4/+4 by adding water
`instead of Triton, but less than 5% of the insulin antigen was
`associated with the membrane pellet. These proteins appear
`to be cytoplasmic by these tests, but we cannot eliminate the
`possibility of some recondite membrane interaction (such as an
`interaction with a protein transport channel).
`
`DISCUSSION
`
`These experiments examined various hybrid signal sequences
`fused to rat proinsulin. Table 1 shows that, in four different
`constructions, 90% of the proinsulin appeared in the periplasm
`when attached either to a bacterial or to a preinsulin signal se-
`quence. In contrast, fusions containing short fragments of the
`signal sequence did not appear in the periplasm. Neither the
`first four amino acids of penicillinase nor two hybrid prese-
`quences formed from the first half of the bacterial signal and
`the last third of the rat signal directed secretion. These fusions
`lack hydrophobic cores, and the last two have added a charged
`amino acid to that region of the presequence. We conclude that
`the signal sequence is essential for secretion; the information
`for secretion cannot reside in the proinsulin moiety alone.
`Furthermore, because either a prokaryotic or a eukaryotic
`signal suffices, the transport mechanism must recognize some
`very general (and very ancient) aspect of structure.
`The signal sequence is not sufficient for transport; other re-
`gions of the protein have some role. Fusions of )3-galactosidase
`to the signal sequence portion of various secreted bacterial
`proteins have yielded hybrids that are membrane-bound or
`remain cytoplasmic (24). However, 3-galactosidase is not a
`secreted protein, and its structure may render transport im-
`possible. Several of our constructions contain the complete
`penicillinase signal attached to a normally secreted protein but
`move only 50% of the proinsulin to the periplasm. These fusions
`contain an extra arginine shortly after the end of the signal se-
`quence; they are transported but with a reduced efficiency. We
`attempted to determine whether or not such partially trans-
`ported proteins were trapped in the membrane. Sonication
`released them, so they are not tightly bound, but more subtle
`experiments, such as proteolytic digestion of the spheroplasts,
`are required to rule out other membrane associations and to
`show definitively that these untransported molecules are
`cytoplasmic.
`In these constructions, a few amino acids from the beginning
`of the penicillinase sequence (and always the fMet) are fused
`to the rat preproinsulin sequence, which lacks its first three
`amino acids. Clearly, these first few amino acids do not domi-
`nate the transport. They are not sufficient in themselves (as
`shown by i4/ +4), and an examination of the 33 (4) available
`pre-sequences reveals the presence in this region of every amino
`acid except tyrosine.
`The striking finding is that both a bacterial and a eukaryotic
`sequence serve to direct efficient transport; this suggests that
`the two presequences play a similar and interchangeable role.
`Fraser and Bruce (25) demonstrated that another secreted eu-
`karyotic protein, ovalbumin, is also transported (50%) when
`cloned and expressed in bacteria. Ovalbumin is unique: it does
`not have an amino-terminal hydrophobic pre-sequence but it
`
`p280.CB15
`
`i25/-7
`
`PR13
`
`p279.CB16
`
`i24/-7
`
`PR13
`
`p241.CBl5
`
`i12/-7
`
`PR13
`
`p234.CB16
`
`i9/-7
`
`PR13
`
`p2l8.47
`
`i4/+4
`
`FMA10
`
`—
`136
`66
`—
`196
`201
`
`—
`(18)
`—
`(34)
`(18)
`—
`(1)
`(1)
`
`—
`132
`77
`—
`212
`192
`
`—
`191
`-—
`342
`176
`—
`10
`15
`
`288 —
`— 51
`— 46
`446 —
`— 48
`— 51
`
`207 —
`— <9
`346 —
`— <9
`— <9
`10 —
`— <9
`— <6
`
`Collected data from radioirnmunoassays. Insulin antigen content:
`P, of the periplasm; C/M, of the cytoplasmic/membrane fraction; WC,
`of the whole cell lysate. The percentage antigen in the periplasmic
`space (% PS) is 100-[P + (P + C/M)]. Data in parentheses are ex-
`trapolated to the 50% inhibition point.
`* Stephanie Broome, personal communication.
`I Ref. 7.
`
`rentheses in Table 1, represent rough maximal estimates. In an
`attempt to increase the yield, we moved some of the construc-
`tions into a polynucleotide phosphorylase-negative strain, E.
`coli PR13 (11), which carries the pnp 13 mutation. In a fresh
`construction, the total number of insulin molecules in each cell
`was more than 10-15 times higher than that found in other host
`strains. However, that high level expression (for example, that
`in i27/ +4, construction p287/47) was lost over time. Efforts
`to maintain stable high level production in the pnp host have
`not yet been successful; we think it likely that the plasmid copy
`number is elevated in these mutants and gradually decreases
`over time.
`
`Table 2 shows that hybrid molecules from four different
`constructions are transported efficiently into the periplasmic
`space: proinsulin fused at four amino acids after the entire
`penicillinase signal sequence (i27/+4), preproinsulin fused at
`two amino acids after the entire penicillinase signal (i25/ -12),
`preproinsulin fused to the first half of the penicillinase signal
`(il2/-21), and preproinsulin fused to just the first four amino
`acids of the penicillinase signal (i4/ -21). In contrast, three
`constructions with fragmentary signal sequences did not secrete
`the insulin antigen. Less than 10% of the material from i12/-7,
`i9/-7, and i4/+4 appeared in the periplasm. Furthermore, the
`
`

`
`Biochemistry: Talmadge et al.
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`3373
`
`Table 2. Amino sequence of hybrid signal sequences and summary of transport data
`
`Pen‘
`i27/+4
`i25/-21
`i12/-21
`i4/-21
`i25/-7
`i24/-7
`
`MSIQHFRVALIPFFAAFCLPVFA
`MSIQHFRVALIPFFAAFCLPVFA
`MSIQHFRVALIPFFAAFCLPVFA
`MSIQHFRVALIP
`MSIQ
`MSIQHFRVALIPFFAAFCLPVFA
`MSIQHFRVALIPFFAAFCLPVFA
`
`MSIQHFRVALIP
`il2/-7,7
`MSIQHFRVA
`i9/-7
`MSIQ
`i4/+4
`Preproinsulin
`
`J ~ HPETLVKL .
`HPET
`HP
`
`HP
`H
`
`.
`.
`. . .
`AAGGGGGG
`LQGGGGG
`LQGGGGG
`AAAG
`LQR
`RCS
`
`WMRF1.PLl.AL1.VLWEPKPAQA
`WMRFLPLLALLVLWEPKPAQA
`WMRFLPLLALLVLWEPKPAQA
`EPKPAQA
`EPKPAQA
`
`LQR
`RCS
`AAGGGGGG
`MALWMRFLPLLALLVLWEPKPAQA
`
`EPKPAQA
`EPKPAQA
`
`1
`
`. >90%
`QHLC. .
`FVKQHLC. .. >90%
`FVKQHLC. .
`. >90%
`FVKQHLC. .
`. >90%
`FVKQHLC. .
`.
`50%
`FVKQHLC. .
`.
`50%
`
`FVKQHLC. .
`FVKQHLC. .
`QHLC. .
`FVKQHLC. .
`
`. <l0%
`. <l0%
`. <l0%
`.
`
`Each sequence begins at the penicillinase fMet and ends at amino acid 7 of proinsulin. Each line represents one continuous sequence which
`has been grouped to emphasize similarities and differences as follows: first group, penicillinase signal sequence amino acids; second group, matured
`penicillinase amino acids; third group, amino acids created by the inserted Pst linker (italics) or by poly(G,C) tailing (glycines); fourth group,
`preproinsulin signal sequence amino acids; fifth group, matured proinsulin amino acids through amino acid 7. The arrows above the prepenicillinase
`and preproinsulin sequences indicate sites of cleavage for maturation. The sequence for prepenicillinase is from refs. 9 and 10; the sequence
`for preproinsulin is from refs. 7 and 22. A, Ala; R, Arg; C, Cys; Q, Gln; E, Glu; G, Gly; H, His; 1, Ile; L, Leu; K, Lys; M, Met; F, Phe; P, Pro; S, Ser;
`T, Thr; W, Trp; V, Val.
`* Penicillinase
`
`may contain an internal hydrophobic region that serves as the
`signal (26). Whatever the process of secretion may be in this
`case, again the bacterial and eukaryotic mechanisms respond
`similarly. We know that cleavage of the signal sequence is not
`essential to transport because a mutation in the signal sequence
`blocks the cleavage of the pre-sequence yet permits 45% of the
`protein to appear in the outer membrane (27). The unifying
`picture is that a hydrophobic “core" sequence, somewhere in
`the protein, serves to attach the protein to an element that leads
`to the passage of the protein across the membrane. The cleavage
`of a pre-sequence then might be related solely to the overall
`efficiency, and irreversibility, of secretion.
`The similarity in behaviors of the prokaryotic and eukaryotic
`signal sequences, which must reflect some underlying similarity
`in structure, raises the question of whether the eukaryotic se-
`quence is processed correctly in bacteria. As we shall show next
`month, it is.
`
`We thank Peter Lomedico, Roger Brent, Marjorie Russel, and
`Stephanie Broome for helpful discussions We thank Stephanie Broome
`and Ann Forsham for generously sharing antibody and Horace Gray
`and Stewart Scherer for the gift of enzymes. We thank Jeremy Knowles
`for being particular. This work was supported in part by National In-
`stitutes of Health Grant AM 21240. ’S._S. was supported by a Damon
`Runyon Fellowship. W.G. is an American Cancer Society Professor
`of Molecular Biology.
`
`1. Palmiter, R. D., Gagnon, I. & Walsh, K. (1978) Proc. Natl. Acad.
`Sci. USA 75, 94-98.
`2. Milstein, C., Brownlee, G. G., Harrison, T. M. & Matthews, M.
`B. (1972) Nature (London) New Biol. 239, 117-120.
`3. Blobel, G., Walter, P., Chang, C. N., Goldman, B. M., Erikson,
`A. H. & Lingappa, V. R. (1979) Symp. Soc. Exp. Biol. 33,
`9-36.
`
`5'-":5
`
`Blobel, G. & Dobberstein, B. (1975) ]. Cell Biol. 67, 835-851.
`Emr, S. D., Hedgpeth, ]., Clement, J. M., Silhary, T. ]. &
`Hofnung, M. (1980) Nature (London) 284, 82-85.
`6. Bedoulle, H., Bassford, P., Fowler, A., Zabin, 1., Beckwith, J. &
`Hofnung, M. (1980) Nature (London) 284, 78-81.
`
`7. Villa-Komaroff, L., Efstradiatis, A., Broome, S., Lomedico, P.,
`Tizard, 11., Naber, S. P., Chick, W. L. & Gilbert, W. (1978) Proc.
`Natl. Acad. Sci. USA 75, 3727-3731.
`8. Bolivar, F., Rodriguez, R. L., Greene, P. ]., Betlach, M. C.,
`Heyneker, H. L., Boyer, H. W., Crossa, ]. H. & Falkow, S. (1977)
`Gene 2, 95-113.
`9. Sutcliffe, C. (1978) Proc. Natl. Acad. Sci. USA 75, 3737-3741.
`10. Ambler, R. P. & Scott, C. K. (1978) Proc. Natl. Acad. Sci. USA
`75, 3732-3736.
`11. Reiner, A. (1969) ]. Bacterial. 97, 1522-1523.
`12. Clewell, D. B. (1972) Bacteriol. 110, 667-676.
`13. Zasloff, M., Cinder, G. D. & Felsenfeld, G. (1978) Nucleic Acids
`Res. 5, 1139-1151.
`14. Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol. 65,
`499-560.
`15. Backman, K., Ptashne, M. & Gilbert, W. (1976) Proc. Natl. Awd.
`Sci. USA 73, 4174-4178.
`16. Mandell, M. & Higa, A. (1970) Mol. Biol. 53, 159-162.
`17. Legerski, R. ]., Hodnett, ]. L. & Gray, H. B. (1978) Nucleic Acids
`Res. 5, 1445-1464.
`18. Broome, S. & Gilbert, W. (1978) Proc. Natl. Acad. Sci. USA 75,
`2746-2749.
`
`19. Makula, D. R., Vichnuk, D., Wright, P. H., Sussman, K. E. & Yu,
`P. L. (1969) Diabetes 18, 660-689.
`20. Osborne, M. I., Gander, E., Parisi, E. & Carson, (1972) Biol.
`Chem. 247, 3962-3972.
`21. Englund, P. (1972) Mol. Biol. 66, 209-224.
`22. Lomedico, P., Rosenthal, N., Efstradiatis, A., Gilbert, W., Kol-
`odner, R. & Tizard, R. (1979) Cell 18, 545-558.
`23. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold
`Spring Harbor Laboratory, Cold Spring Harbor, New York), pp.
`352-355.
`24. Bassford, P., Silhavy, T. 1. & Beckwith, J. (1980) 1. Bacteriol. 139,
`19-31.
`25. Fraser, T. & Bruce, B. I. (1978) Proc. Natl. Acad. Sci. USA 75,
`5936-5940.
`26. Lingappa, V. R., Lingappa, I. R. & Blobel, G. (1979) Nature
`(London) 281, 117-121.
`27. Lin, J. J. C., Kanazawa, H., Ozols, I. & Wu, H. C. (1978) Proc.
`Natl. Acad. Sci. USA 75, 4891-4895.