Gene. 121 (1992) 111-119 0 1992 Elsevier Science Publishers
GENE
B.V. All rights reserved.
111
037X-l 119/92/$05.00
06726
Secretion of human blood coagulation lipolyt ica (Yeast expression
vectors;
heterologous
Ckcile Tharaud a, Anne-Marie Ii Lahorutoire Received
de GCnPtique INRA-CNRS
by J.K.C.
Knowles:
rerouting;
transglutaminase;
clotting
cascade)
Ribet a, Claude Costes b and Claude Gaillardin a
and’
31 January
secretion;
factor XIIIa by the yeast Yarrowia
Laboratoire
de Chimie Biologique, Institut National Agronomique
1992; Revised/Accepted:
22 May/9 June 1992; Received
Paris-Grignon,
at publishers:
78 850 Thiverval Grignon, France
7 July 1992
SUMMARY
The industrial yeast, Yarrowia lipolytica, secretes high yields of an alkaline extracellular protease (AEP), which is synthesized as a preproprotein encoded by the XPR2 gene. We investigated the possibility of using this system for the secretion of human coagulation factor XIII subunit a (FXIIIa). This protein is naturally secreted in the plasma by an unknown, signal peptide-independent mechanism and has so far been found to be nonsecretable in yeast. We have designed six hybrid genes encoding fusion proteins between increasing portions of the AEP preprodomain and the precursor or mature forms of FXIIIa. All constructs directed translocation of the FXIIIa precursor into the endoplasmic reticulum. Transport of the translocated and core-glycosylated hybrid precursor to the Golgi apparatus appeared to be strongly rate limiting, and most of the precursors appeared to be partially proteolysed. One of these constructs directed the extracellular secretion of a low amount of hyperglycosylated FXIIIa. These results indicate that fusion to the yeast AEP signal peptide and dipeptide stretch allows FXIIIa to be translocated, albeit inefficiently, through the endoplasmic reticulum and to follow a classical secretory transit.
INTRODUCTION
Factor XIIIa (FXIIIa) is the catalytic part of human blood coagulation factor XIII, the last zymogen to become
Correspondence
PG-CBAI,
to: Dr.
C. Gaillardin,
78 850 Thiverval
Fax(33-1)
Laboratoire
Grignon,
France.
de Gtnttique,
30 81 54 57. A, absorbance
Abbreviations: extracellular
protease;
(1 cm); aa, amino acid(s);
Ap, ampicillin;
bp, base pair(s);
linked immunosorbent assay; EndoH, endoglycosidase XIII; FXIIIa, FXIII subunit a; FXIIIa, gene encoding terferon;
INA
Te1.(33-1) 30 81 54 52;
kb, kilobase
ribonucleotide;
PAGE,
merase
reaction;
chain
or 1000 bp; nt, nucleotide(s); polyacrylamide-gel PMSF,
AEP, alkaline
ELISA,
oligo, oligodeoxy-
electrophoresis;
phenylmethylsulfonyl
enzyme-
H; FXIII, factor FXIIIa; IFN, inPCR, fluoride;
polyS.,
Saccharomyces;
activator; extract/2”, 6.8.
SDS, sodium dodecyl sulfate; tPA, tissue plasminogen XPRZ, gene encoding AEP; Y.. Yarrowia; YPDm, I”,; yeast glucosej59,
Difco Bactopeptone/50
mM phosphate
buffer pH
activated during the clotting cascade. Thrombin-activated FXIIIa (FXIIIa*) is a transglutaminase (EC 2.3.2.13) which promotes the stabilization of the fibrin clot: it catalyzes the formation of covalent intermolecular y-glutamyl&-lysine bonds between fibrin CIor y chains (McDonagh, 1987). Factor XIII congenital disorders (Girolami et al., 1991) or partial acquired deficiencies in patients suffering from severe illnesses will provoke lifelong abnormal bleeding tendencies, lethal intracranial hemorrhages, abnormal wound healing, spontaneous abortion. Up to date, replacement therapy consists of monthly injections of placental FXIIIa concentrate (Rodeghiero et al., 1991). Since it is free of blood contaminants, recombinant FXIIIa may be valuable both as a substitution product and as a component of surgical glues (Lindsey et al., 1990). Recombinant FXIIIa would also allow food texturation during fermentation processes, through reticulation of proteins such as
112 myosin, caseins, or soya globulins (Traore and Meunier, 1991) and in this regard, secretion of even low levels of FXIIIa by a suitable microorganism may be useful. Authentic recombinant FXIIIa has already been obtained as a cytoplasmic protein in the yeast S. cerevisiae (Bishop et al., 1990b; Jagadeeswaran and Haas, 1990; Rinas et al., 1990), but attempts to have it secreted by yeast remained unsuccessful until now (Bishop et al., 1990a). Several aspects of FXIIIa biosynthesis remain obscure, in particular its secretion process. The FXIIIa homodimer does not reach circulating plasma after a classical secretory transit (Muesch et al., 1990). Molecular analysis revealed that its primary sequence shows no N-terminal nor internal signal sequence; that no carbohydrate is attached to any of its six potential N-glycosylation sites; that its N-terminal Ser residue is acetylated; and that none of its nine Cys residues is involved in a disulfide bond (Grundmann et al., 1986; Ichinose et al., 1986; Takahashi et al., 1986). These features are consistent with a cytoplasmic localization, and do not provide any clue on the natural secretion pathway of factor XIIIa. The yeast Y. Zipolytica presents a great potential for heterologous protein secretion. It naturally secretes an alkaline extracellular protease (AEP), encoded by the XPR2 gene, at levels up to l-2 mg/ml at high cell densities, and has been used to direct successfully the production and secretion of bovine prochymosin, S.cerevisiae invertase, porcine al-interferon, and human tPA (Buckholz and Gleeson, 1991; for a review see Heslot, 1990). AEP is initially synthesized as a large precursor with a 1%aa signal peptide, a stretch of nine X-Ala/X-Pro dipeptides (Ala-Pro-LeuAla-Ala-Pro-Ala-Pro-Ala-Pro-Asp-Ala-Ala-Pro-Ala-AlaVal-Pro), a 122-aa propeptide followed by a Kex2-like cleavage site, and a mature part (Matoba and Ogrydziak, 1989). We show here that fusion to the pre-(X.Ala/X.Pro) domain of AEP allows the secretion of FXIIIa into the growth medium.
RESULTS
AND DISCUSSION
(a) FXIIIa cDNA FXIIIa cDNA human placental Alto, CA) with
cloning clones were obtained by screening a lgtl 1 library (Clontech Laboratories Inc. Palo three 32P-labeled oligo probes designed
a All oligos were synthesized under each oligo indicates
with a Cyclone whether
DNA synthesizer
its sequence
reproduces
(Biosearch). the mRNA-like
from published nt sequence of FXIZZa cDNA (Grundmann et al., 1986; Ichinose et al., 1986) (for oligo sequences, see Table 1). The frequency of positive clones (3 out of 150 000 phages screened) was consistent with published data concerning immunoscreening of this expression library (Ichinose et al., 1986). The three inserts were subcloned in pBR322 as BamHI fragments; restriction analysis and comparison with published data showed that the longest insert was nearly full length. Sequence analysis revealed that the cloned FXZIZa cDNA insert lacked the 5’ noncoding region and the first five nt encoding the start codon and part of the N-terminal Ser’ codon. The rest of the sequence up to the RglII site (nt = 2814) was strictly identical to that published by Ichinose et al. (1986). (b) XPR2-FXZZZahybrid genes design and assembly; the (pre-X.Ala/X.Pro),,, directs the secretion of FXIIIa To direct the secretion of FXIIIa via the secretory pathway, we fused increasing lengths of the XPR2 region encoding the AEP prepropeptide to cDNA fragments encoding either the zymogenic form of FXIIIa or the activated form of FXIIIa, FXIIIa*. Since the 3’-untranslated region of FXZIZa cDNA was seen to affect negatively protein production in Y. lipolytica (not shown) as already reported in S. cerevisiae (Bishop et al., 1990a; Broker et al., 1991) it was deleted from the cDNA clone. The reconstituted shortened cDNA was inserted downstream from the XPR2 promoter and preprosequences, and upstream from a minimal XPR2 terminator sequence to generate pINA (see Fig. 1). Deletions in the preproregion were generated in vitro and inserted in pINA between the BglI or AvaI, and SmaI sites (Fig. 1) so as to abut the N-terminus of FXIIIa to the AEP signal sequence (Ala15; p-13a), to the dipeptide stretch (Pro’3; pD-13a), or to the entire prosequence (Arg15’; pDP-13a and pDP-13a*). The cDNA fragment encoding activated FXIIIa (FXIIIa*) was only fused behind the entire XPR2 preprosequence, so that no mature enzyme should be generated in the secretory pathway before the late Golgi apparatus, where KexZ-like proteolytic processing was expected to occur. All junctions between the XPR2 preprosequences and FXlIIa cDNA were generated by PCR so as to generate precise fusions and to reconstitute the 5’-terminal coding sequence of the FXZIIa cDNA (see Table I for the fusion sequences). A few silent mutations Their sequences (rightward)
are given from 5’ to 3 ’ , except for LA2 and LA4. The arrow
or complementary
(leftward)
strand
of the corresponding
gene.
(1) indicates the fusion site between the XPR2 and FXIIIa sequences. The aa sequences are indicated under each fusion oligo, from N- to C-terminal. Small letters indicate mutated nucleotides. Screeningprobes. Oligos were derived from the FXIIIu published sequences (see section a). Linkers. LAI and LA2 were used to construct pINA (section b and Fig. l), LA3 and LA4 to generate the pDP-13a* fusion (section b and Fig. 2). LA 1 and LA2 inserted six mutations
in the XPR2 sequence
in order to create
a Sfl site; three of those introduced
an Ala codon
replacing
Asn”‘.
The original
AsnlSJ
was restored by cloning the LA3-LA4 linker into pINA cut by Sfi + &a1 to create the pDP- 13a* fusion (see Fig. 2). PCR primer.\. The primers V.VC used to create the different fusions shown in Fig. 2. TC15 and TC16 were designed to bring additional restriction sites (Sfil and Yhol).
113 TABLE
I
Oligos used for hybrid genes construction
Screening
CT1 GAC +
ACC
*
AGC AAA
AAC
CCA GCA TTG
GGC
(FXlllo....Gly”‘)
probes CT2 GGA GGA GAT +
CT3 GAC +
GGC ATG
CCA GGG CCC
ACT CAT
GCA CAC TTC
I-W
CCA CTG CAC
CC8 GCc TCT
TCg 8cc GCC AAG
CGA
GQ
AGc ceg CGG TTC
GCT
CGe AGA
(XPRZ .. Lys CT AAT GA AGA TTA
A
Tz
II Ii
KpnI
CCCGGGTGGACGTCTAGAGGTAC LA2 c
GGGCCCACCTGCAGATCTC
Argls’//
polylinker)
GCC AAG C CGG TTC
G
LA4
(XPR2....Lys’““)
PCR
XbaI
SmaI
SfiI ATT
C TAA
LA3 +
ACT TAC
(FXllh....Val6’0)
LA1 CC GAG -
ATT
(FXIIh...Tyr’s’)
A vaI
Linkers
ATG GAT
-
va1
CTC CCC
ATT
CCT
GCT TCT
TCT
AAT
GCC AAG
CGA
IlTCt
GA8 ACT TCC AGG
ACC GCC ‘ITT
Lys Arg’“‘//Ser~. . . FXllln)
(XPRZ...
primers
EglI Tz
ACT
GCC GTT CTG GCC GCT CCC CTG GCC (XPR2..Aln’9) GGC
TC5 TCC +
AAA
TC6 CTC +
ACT a
TC8 AAG +
AAA
GGT CCT GGA AGT
cTC get
GTT
CTG GCC
IiTCt
II AGG
CAC AGC
AGC AGG
GGC AGC
ATC
Pro’=// Se+. . . FXIIIa)
(XPRZ......
GA8 ACT TCC AGG
ACC GCC TTT
GGA
(XPR2 . . . . . Al~~“//Serl...FXllI~) CTC TTG
CAG
GTT GAC GC (nt 200 of FXIIIa
sequence,
reverse complementary
TC12 GGCAAACTATCTGTTAATTGCIITCA CAT (XPR2 Terminatorl/Stop..EXIllo)
CGA
AGG TCG
(reverse
TCT
strand) TTG
complementary
AAT strand)
XmnI TC13 CCA
ATG
AA-
ATG T’I% CGT (FXIIIa...Arg6s*)
BglII ‘24
No11
GCTGCAAGATCTGCGGCCGC
GCCACCTACAAGCCAGA
(nt 2382 of XPR2
reverse complementary BglI TClS -
CTC
Tz6
CTC
ACT
GCC GTT
SfiI CTG m
GCc CCC CTG CCC/l
sequence,
strand)
XbaI TCt GA8 ACT
TCt
ACC
GCC TTT
(XPRZ ..,, A Ials// Ser*. ,. FXlIla) BglI ACT
U
XbaI
GTT CTG CCC
GCT CCC //TCt GA8 ACT TCt
(XPR2...Pro~~llSer~...FXIlIa)
ACC GCC TTT
GGA
GGA
GGA
114 EcoRI
EcoRI Hind111
Hind111 I I he1
AvaI StiI SmaI XbaI KpnI
pINA (8.7 kb)
Eco
Minimal XPRZ terminator
Sal1
Not1
ti
EcoRI
Hind111
m
XPR2 promoter AEP prodomain
El
FXIIIa domain
0
Marker gene
pINA (7.1 kb)
FXIIIa* Fig. 1. Construction EcoRl-BglII a pBR322 fragment
of FXIIIa
fragment derivative
pINA502.
We cloned
with a 3’ Not1 restriction (Nicaud
This construct
was then fused downstream Briefly, pINA
sequence
plasmid
fragment
was cloned
placental
and 534 bp of the 3’ flanking Separately,
[from XmnI (nt = 2120) to stop codon]
into pINA366.
from the XPR2 promoter
sequence, replacing
library
(see section
a). An
region, Lvas subcloned
we synthesized
into
by PCR a XmnI-Bg/II
fused to a short XPR2 terminator
to Yon and Fried (1989), using pINA
as the polymerization
(130 bp
malrix for the
with TC12 as fusion oligo. and TC I3 and TC 14 the 3’ untranslated
and preprosequences
all the 5’XPRZ prepro coding sequences,
contains
from a igt I1 human
was called pINA366.
et al., 1989) as the matrix for XPR2 3’-flanking
(see Table I). The resulting
elsewhere).
fragment
except the two first codons
site. This was done according
PCR primers
be described
an FXIIIa cDNA
sequence,
a Bg/II site in place of PvuII. The resulting
FXIIIa 3’ coding region, pINA as external
vectors.
the last 160 bp of the FXIIIa coding
which carried
from stop codon)
secretion
the entire FXIIIa coding
carrying carrying
BamHI
region of F,%‘IIIu. This yielded
derived from the expression
the LAl-LA2
oligo adaptor
vector pINA
(to
(see Table I) inserted at the AwI
site (nt 2537) present at the end of the AEP prosequence, encoding
the AEP prodomain,
tween the EcoRI-SmrrI
fragment
with one substituted from pINA476,
followed by a synthetic XPR2 terminator. This linker reconstitutes the 3’-end of the XPR2 sequence aa (Asn ‘s4); it also brings three unique cloning sites: .SfiI, SmaI, and XhaI. A three way ligation bc-
SmaI-BglII
fragment
from pINA50,2
and BglII-EcoRI
fragment
from pINA
yielded pINA503.
This
plasmid was then used as the recipient for all the PCR synthesized XPRZ-FXIIIu fusion fragments described in Table 1, to generate a series of plasmids which carried the hybrid cassettes represented on Fig. 2, under the transcriptional control of the complete XPR2 promoter and of a minimal XPR2 terminator and excisable by a M/u1 + Nor1 digestion. To insert these cassettes into yeast integrative or replicative vectors carrying selcctablc markers. wc constructed pBR322 derivatives carrying a Not1 linker al the PvuII site, a Mu1 linker at the BumHI site, and the 5’ upstream region of XPR2 as a HindIII-M/u1
fragment
inserted between Hind111 and MuI. As required, we introduced
et al., 1989) or LEU2 as an EcoRI fragment sued from pINA plasmids,
(Fournier
from pINA
(Gaillardin
et al.. 1991): see, for example,
so that each of the fusion genes could be expressed
3’ sequences flanking the XPR2 coding scquenccs arrow, marker genes as large dotted arrows.
are depicted
pINA437.
URA3 between the EcoRI and Hind111 sites (from pINA 156; Nicaud
and Ribet, 1987) and/or
finally the I’.
The M[uI-Not1 XPRZ-FXIIIa hybrid
in an integrative
or replicativc
context,
as thin black boxes, preprosequences
together
lipo&ica
cassettes
ARSIB as a RUMHI fragment were cloned
is-
into each of these
with LEU2 or URA3 markers.
The 5’ and
as thick black boxes, FXIIIu sequences as an open
115 were introduced in the FXIIIa 5’ coding region, in order to establish a more favorable codon bias at the fusion site (Heslot, 1990). After insertion into pINA503, the hybrid constructs were excised as MuI-Not1 cassettes, and were transferred into yeast transforming vectors as described in Fig. 1. The resulting plasmids were used to transform Pold, a Y. lipolytica uru3leu2xpr2 strain, by the lithium acetate procedure (Gaillardin and Ribet, 1987). This strain (to be described elsewhere) is a derivative of the industrial strain W29 (ATCC20460) where non-reverting mutations were created by transformation. Integrative plasmids were linearized by Mu1 digestion to target the constructs to the XPR2 chro-
Em
AEP signal
ml
AEP X-Ala/X-Pro
a3
AEP prodomain
sequence
I
FXIII prodomain
I
AEP or FXIIIa
+
Known
Puten,ia,
+
Fig. 2. Schematic
stretch
mature domain
glycnsylation
site(s)
of the XPR2-FXIIIa
representation
The first two boxes represent the sequences .4EP precursor (pDP-AEP) and the zymogen indicate
N-glycosylation
cellular transit osylated
in human
pDP-13a,
plasma
while none of the FXIIIa
samples
the constructs
pDP-13a*,
respectively, the (13a). Diamonds
sites: the AEP site is glycosylated
(large diamond),
boxes represent
hybrid constructs.
encoding, of FXIIIa
pD-13a,
(small diamonds).
encoding
AEP-FXIIIa
p.2d-13a,
p.ld-l3a,
during intrasites are glyc-
The following secretory and p-13a,
six
fusions: where p
stands for the AEP slgnal sequence, D for the AEP X.Ala or X.Pro dipeptidc stretch, 2d for the first two dipeptides of the former stretch, Id for the first one, P for the AEP prodomain, and 13a* for the sequence
encoding
13a for the FXIIIa
activated
FXIIIa.
sent the oligoa which were used for the construction
sequence,
The arrows
repre-
of the hybrid genes.
LA oligos form a fusion linker, TC oligos were used as PCR primers (see Table 1). The pDP-13a, were generated polymerization
p.2d-13a.
and p-13a junction
and (TC6 + TC8) respectively. and
as the pairs of primers.
Primers were used
1 1 min at 72’C were carried out. The gent fusions beof
I PM. Twenty cycles of 0.25 min at 94’C,
tween the 3’ end of FXIlIrr
and XPR2 terminator
pD-13a fusion were obtained
according
matrices
fragments
using IO ng of pINA (see legend to Fig. I) as the matrix and (TC2 + TC8). (TC15 + TC8), (TC16 + TC8)
at a final concentration min at 55’C
p.ld-13a
PCR reaction.
(see Fig. 1) and the
to FXIIIu downstream
SnzaI site (nt 196). so that this site was included
from the
in each of the XPR2-
FX1lltr joins. TC2 carried
TCl6
and TC6 carried
was thus cloned while pD-13a.
the XPR2 .4wI site (nt 2537). and TC4, TC15, the XPR2 Bg/I site (nt 2138). The pDP-13a joint
into pINA
p.2d-13a,
p.ld-l3a.
(c) Influence of the AEP signal peptide on FXIIla secretion As mentioned above, the hybrid gene encoding p-13a did not direct the secretion of FXIIIa, whereas that encoding pD-13a did. We hypothesized that fusing of the FXIIIa hydrophilic N-terminus right after the last residue of the AEP signal peptide (Ala”) in p-13a might interfere with signal peptide recognition and/or cleavage. To test this
to Yon and Fried (1989) by a two
TC5 was used as the fusion oligo and TC4 and
TC8, as external primers. TC8 hybridized
mosomal locus of the recipient strain. Replicative plasmids carried the Y. lipolytica autonomously replicating sequence ARS18 which, exhibits centromeric functions and results in stable, low-copy-number plasmids (Fournier et al., 1991). All transformants were checked either by Southern blot analysis of chromosomal DNA for integrants, or by restriction analysis of reextracted plasmids for replicative transformants (Fournier et al., 1991). In order to identify the constructs which could direct the secretion of FXIIIa into the growth medium of transformed yeasts, stable Pold integrants were grown under XPR2 inducing conditions (YPDm medium) at 18°C. Samples were collected during early stationary phase when AEP production in a wild-type strain reaches a plateau (Nicaud et al., 1989a). Supernatants were concentrated by ultrafiltration and aliquots were deglycosylated using EndoH. Crude and deglycosylated samples were subjected to SDSPAGE and immunoblotting using anti-FXIIIa antibodies. Results presented in Fig. 3 show that the hybrid gene encoded by pD 13a directed the secretion of an immunoreactive product (compare lanes 1 and 4, 2 and 5). In contrast, none of the other constructs led to the secretion of an FXIIIa-immunoreactive material (data not shown). Results concerning the secretion of pD 13a protein are dealt with in section e.
(see Fig. 4) as an AvaI-XmuI
fragment,
and p-13a joints were cloned as BgII-
XnzaI fragments. analysed
After fragment
by restriction
figure is not drawn alternance dipeptides;
enzyme
insertion
into pINA503,
digestion
and dideoxy
to scale. The AEP signal peptide
all fusions were sequencing.
This
is 15 aa long; the
of black and white boxes represents the AEP series of nine the AEP prodomain contains 124 aa, the AEP mature part 297
aa: the FXIIIa part, 694 aa.
activation
peptide contains
37 aa and the FXIIIa
mature
116
A
pD-13a
T
T
ARS18
13a
---II
_ -------
_
+
+
-
kDa
lP;,
13a kDa
v~~~~-
+
~~~,#1
B
“”
-
91.4
-
69
-
46
I
200
‘ii
4,
-
12
97.4
34
B T
-69
56
7
pF3;-
PD13a
13a
8
9
10
11
pDP13a*
La
kDa --B---B----
-
123
4
Fig. 3. FXIIla natants
secretion.
(pD-13a) or from a replicative control
strain
(T). Crude
expressing
(- ) and
the analysis
pD-13a
(pD-13a/ARS)
for each strain. Placental
-
200
-
97.4
-
69
-
46
_.“I-‘--
,,__
_^_
67
The figure only presents strains
from the Leu’
compared
5
-
plasmid
EndoH-treated FXIIIa
of super-
from an integrated and from the Leu + (+ ) samples
were
(25 ng) purified from Hoechst
Fibrogammine (Traore and Meunier, 1991) was loaded as a control (13a). All strains were grown at 18°C on YPDm inducing medium as previously described (Fabre et al., 1991). Samples were collected ture, at a cell density of &,,,“,,, of 18 units, and PMSF centration),
and aprotinin
centrifuged
at 12000 xg for 10 min at 4°C.
EndoH
treatment.
ice-cooled, PMSFjS
and
l/l000
v/v) were added.
Samples were adjusted then
adjusted
mM Na.azide.
37°C after the addition or deglycosylated through
(Sigma;
to 50 mM
Millipore Ultrafree-MC
on 0. I “/b SDS/7.5%
polyacrylamide
teins were transferred membranes Diagnostica; anti-goat
to nitrocellulose
were probed I:1000
antibody
pH
5.512 mM
(Genzyme).
Then 400 ~1 crude
were submitted
to ultrafiltration concentrated
sample buffer. Aliquots were subsequently
(Schleicher
and an alkaline
(50 ~1,
fractionated
gels. After electrophoresis,
with a goat polyclonal
dilution)
(Biosys;
Na,.citrate
filter units 30000 NMWL,
to 200 ~1 culture supernatant)
to
out for 2 x 8 h at
to 50 pl, and diluted in 50 ~1 2 x Laemmli equivalent
Samples were
were subjected
was carried
of 1 mU of EndoH aliquots
Aliquots
123
to 0.1% SDS, boiled for 3 min,
Deglycosylation
supernatant
after 41 h of cul(2 mM final con-
the pro-
and Schuel), and the anti-FXIII
(American
phosphatase-conjugated
1:800 dilution).
possibility we generated two constructs (p.ld-13a and p.2d-13a, see Fig. 2) carrying one or two X-Ala/X-Pro dipeptide(s) after the signal peptide cleavage site, so as to possibly restore a suitable structure around the cleavage site, without introducing the whole dipeptide stretch. According to von Heijne’s rule (von Heijne, 1986) signal peptidase cleavage was predicted after Alal (not shown). However, none of these extended signal peptides directed secretion of FXIIIa (data not shown).
Fig. 4. Western
4
5
blot analysis
67
8
of FXIlIa
9
10
intracellular
11
accumulation.
We
crude ( - ) and EndoH-treated (+ ) samples of the Leu + conexpressing the trol strain (T), and of the Leu * integrative transformants
compared
constructs FXIIIa
indicated
on top of panel
(25 ng) purified from Hoechst
A and panel
B. (13a), placental
Fibrogammine
(Traore
and Meu-
nier, 1991). Cells were grown
and collected as described in Fig. 3. Cell pellets equivalent to AbOOnmof 4 were lysed in 80 ~11Laemmli sample buffer, heated at 95°C for 5 min, and disrupted by vortexing with 0.3 g of acid-washed
glass beads (0.45 mm). Then 320 ~1 of 1 x sample buffer
were added, and samples were boiled for 5 min; extracts equivalent
to AhOOnmof 0.6 were analysed
as described
issued from cells in the legend to
Fig. 3.
The four strains encoding p-13a, p.ld-13a, p.2d-13a and pD-13a were analysed for their intracellular content of immunoreactive FXIIIa. Crude and EndoH-treated cell extracts were subjected to immunoblotting with anti-FXIIIa antibodies after electrophoresis under denaturing conditions (Fig. 4A). The AEP signal peptide alone (p-13a construct) led to a low production of precursors as compared to pD-13a (Fig. 4A, lanes 3, 4 and 9, 10; Figure 4B, lanes 6, 7. and 8,
117 9). The p-13a products into the endoplasmic
were at least partially translocated reticulum, since treatment with
EndoH resulted in a mobility shift from approx. 100 kDa to approx. 88 kDa. This was consistent with removal of six N-linked core-carbohydrate chains. The slight difference observed between native FXIIIa (84 kDa) and these EndoH treated precursors (88 kDa) was not due to interfering yeast protein (native FXIIIa migration was not disturbed by addition of an equivalent amount of yeast protein, not shown). It may reflect residual N-acetyl glucosamine moieties or O-glycosylation, or retention of (part of) the AEP signal peptide due to impaired signal peptidase cleavage (see above). EndoH-treated cell extracts of the strain expressing the (secreted) pD- 13a construct exhibited a single 92 kDa band, suggesting that the hybrid protein was translocated and core glycosylated, possibly at all six Asn-X-Thr/Ser sites present in the FXIIIa zymogen (see Fig. 2). All sites however did not appear to be recognized and glycosylated at similar rates, since native samples revealed precursors of 102, 106 and 110 kDa (independent experiments, not shown). The different mobility of the deglycosylated precursors accumulated by strains expressing p-13a and pD13a suggested that the latter was not processed by dipeptidyl-aminopeptidase. We also examined the intracellular content of the strains encoding the two intermediate constructs. The p.ld-13a precursor accumulated as a glycosylated but degraded product of 58 kDa, bearing 6 kDa of N-linked core carbohydrate chains (Fig. 4A, lanes 7, 8). We have not elucidated whether this form was entirely translocated inside the endoplasmic reticulum or if it remained partially exposed to a putative cytoplasmic degradation. The expression of p.2d-13a precursor could not be detected (Fig. 4A, lanes 5, 6). Taken together, these results suggested that the signal peptide of AEP alone was able to direct the translocation of FXIIIa into the endoplasmic reticulum, but that the whole dipeptide stretch was important for FXIIIa secretion outside of the cell. (d) Role of the AEP propeptide in the secretion of FXIIIa The construct pDP-13a, with an entire AEP preprosequence, did not direct the secretion of FXIIIa. The immunoblot analysis of cell extracts revealed intracellular accumulation of glycosylated precursors. Before treatment with EndoH, we observed a major polypeptide of 72 kDa, and a minor one of approx. 90 kDa (Fig. 4B, lane 4). Removal of N-linked carbohydrates resulted in a 6-kDa decrease of both precursors yielding two products: a major one migrating faster than native FXIIIa (around 66 kDa), and a minor one with a mobility similar to that of FXIIIa (Fig. 4B, lane 5). This indicated that the precursors issued from pDP- 13a
expression were translocated, core-glycosylated, but that most of them underwent aberrant proteolytic cleavage. The amount of precursors accumulated was lower for pDP-13a* than for pDP-13a (Fig. 4B, compare lanes 4, 5 and 10, 11) but they similarly underwent limited degradation. A single 68-kDa product could be observed in the crude extract, which was reduced to 62 kDa following deglycosylation (vs. 80 kDa for FXIIIa*). The 4-kDa difference between this precursor and the 66 kDa of the major deglycosylated form issued from pDP-13a reflected the absence of the 4-kDa FXIIIa activation peptide. In summary, the presence of the whole AEP proregion within the hybrid precursors seemed to impair secretion and to promote aberrant proteolytic processing. (e) Secreted factor XIIIa As discussed above, the construct encoding pDP-13a was the only one to direct secretion of an immunoreactive product, although the efficiency of this process was low. Western blot analysis of supernatants concentrated by ultrafiltration showed a hardly detectable smear corresponding to a heterogeneous product composed of high M, species, while EndoH-treated samples allowed detection of a diffuse smear around 84 kDa (Fig. 4). Smearing after EndoH treatment may be due partly to the high amount of peptone present in concentrated YPDm samples, although attempts to purify the product could not rule out that the secreted product itself was heterogeneous (see below). We feel it unlikely that the extracellular accumulation of FXIIIIa related products could reflect cell lysis, since no trace of the intracellular form of 92 kDa was recovered outside of the cells. This rather suggested that passage through the Golgi apparatus resulted in secretion of a hyperglycosylated protein. We compared the secretion of FXIIIa-related products by strains carrying the hybrid construct either integrated at the XPR2 locus or on a replicative plasmid, with LEU2 or URA3 as selective markers. A slight increase in the amount of secreted immunoreactive material was observed with replicative plasmids as compared with integrated plasmids (Fig. 3). This reflected the high stability but low copy number (1 or 2 per cell) of plasmids carrying the Y. lipolytica ARS18 sequence (Fournier et al., 1991). No difference was observed in connection with the marker used (not shown). Temperature did not affect the level of FXIIIa secretion (data not shown). In order to quantify the secreted product, crude and deglycosylated supernatant aliquots were subjected to a double site ELISA, using a mouse monoclonal FXIIIaspecific IgG as the capture antibody and a rabbit polyclonal serum as the second antibody (isolation of monoclonal IgG specific for FXIIIa and development of an ELISA assay are to be published elsewhere). This assay indicated a pro-
118
duction of approx. 500-600 ng/ml for integrative or replicative pDP- 13a transformants, whereas control strains or nonsecreting transformants gave a lower than the threshold value of this test (5 ng/ml). In order to purify the hyperglycosylated secreted product away from the peptones, we concentrated a 2.5-ml aliquot of the supernatant from an integrative transformant through ultrafiltration followed by adsorption on concanavalin A Sepharose. The product was then eluted with x-D-methylmannopyranoside, subjected to EndoH treatment, SDS-PAGE, and immunodetection. The resulting immunoblot membrane exhibited a fuzzy band in the 85kDa range. The heterogeneity of the observed product may thus reflect heterogeneity of the secreted FXIIIa, due to limited proteolysis or other modifications such as O-glycosylation. This, however, confirmed that the amount of secreted FXIIIa did not exceed 1 pg/ml culture supernatant and precluded assessment of its transglutaminase activity. (f) Conclusions In this paper, we show that the signal peptide of AEP together with the X.Ala/X.Pro region can direct, albeit inefficiently, the secretion of human factor XIIIa in Y. lipo@tica. (1) Human FXIIIa is naturally secreted by an unknown, signal peptide-independent mechanism (Muesch et al., 1990). It is worthwhile noticing that FXIIIa expressed as a cytoplasmic protein, either in S. cerevisiae (Bishop et al., 1990a) or in Y.l@oLvtica (unpublished) is not spontaneously secreted outside the cells. We show here that FXIIIa can be secreted by the classical secretory pathway of Y. lipolytica, when expressed as an AEP-FXIIIa fusion precursor. To our knowledge this is, along with human interleukin lp (Baldari et al., 1987; Livi et al., 1991) the second example of a secretory protein without a classical signal peptide which has been rerouted through a yeast secretory pathway. Both interleukin l/j’ and FXIIIa underwent aberrant N-linked glycosylation. In the case of FXIIIa, transit through the secretory pathway resulted in additional modifications (perhaps limited proteolysis, O-glycosylation,...) which led to the formation of a heterogeneous product. (2) Whereas secreted FXIIIa was hyperglycosylated (see section e and Fig. 3), no hyperglycosylated form was detected within the cells: if undetected hyperglycosylated polypeptides had been present in the crude extracts, EndoH treatment should have enhanced the immunoblot signal, which was not the case (Fig. 4A, lanes 3, 4 and Fig. 4B, lanes 6, 7). This suggested that a rate-limiting step slowed down the transit to the Golgi apparatus, where outer-chain addition takes place, but that later steps were not rate limiting in the case of the secretable pD-13a construct. Nonsecretable constructs yielded proteins which were par-
tially degraded, probably inside the secretory pathway. This degradation may reflect early proteolysis of the heterologous hybrids or targeting to the vacuole and subsequent degradation. (3) The results presented here indicate that the whole prosequence of the XPR2 gene may not be useful for secreting heterologous proteins: it has no influence on the secretion of porcine IFN or bovine prochymosin (Heslot, 1990) and seems to hamper the secretion of FXIIIa. Interestingly, the prosequence of Kluyveromyces lactis killer toxin was shown to negatively affect the secretion of human serum albumin in S.cerevisiue (Sleep et al., 1990). Prosequences might be generally dispensable for the secretion of heterologous proteins, as already suggested by some reports on the prosequence of S. cerevisiae x-factor (Ernst, 1988). In the case of Y. lipolytica AEP, the prodomain was reported to be crucial for the intracellular transit of the protease precursor (Fabre et al., 1991). However, this function might be comparable to that of an internal chaperone and be specific for the AEP mature part (Fabre et al.. 1992). Whereas the whole prosequence seemed to be unnecessary or even deleterious for FXIIIa secretion, the signal peptide alone was not sufficient for extracellular secretion. Surprisingly, the AEP dipeptide stretch seemed to favor intracellular transit, as indicated by the results presented in section c. We have currently no explanation for the facilitating role of the dipeptide stretch in the secretion of recombinant FXIIIa, whereas its exact role during AEP secretion rcmains hypothetical (Kreil, 1990; Matoba and Ogrydziak, 1989).
ACKNOWLEDGEMENTS
We thank Philippe Joyet for the synthesis of oligos, Feng He for the PCR synthesis of the minimal XPR2 terminator used to construct pINA476, and Emmanuelle Fabre for the construction of pINA302. Special thanks to Christophe d’Enfert for helpful critical reading of the manuscript. Part of this work was supported by SOREDAB (Guyancourt, France).
REFERENCES Baldari, C., Murray, J.A.H., Ghiara, P., Cesareni, G. and Galcotti, C.L.: A novel leader peptide which allows efficient secretion of a fragment of human
interleukin
18 in Saccharwn~re.s
crrevisinr.
EMBO
J. 6
(1987) 229-234. Bishop, P.D., Teller, D.C., Smith, R.A., Lasser, G.W.. Gilbert, T. and Scale, R.L.: Expression, purification, and characterization of human factor
XIII
in Smchurmqces
wrevisim.
1861-1869. Bishop, P.D. and Lasser. G.W.. Trong,
Biochemistry
I.L.. Stcnkamp,
29 (1990a)
R.E. and Tcllcr.
119 D.C.: Human -
recombinant
crystallization
(1990b)
factor XIII from Saccharomyces
and preliminary
x-ray
data.
cerevisiue
J. Biol. Chem.
265
of the human
in Sacchuronzr~es host-vector
blood coagulation
cerevisiae: dependence
systems
and medium
of heterologous
M. and Amann,
protein
factor XIIIa
of the expression
conditions.
levels from
Appl. Microbial.
Bio-
Yeast systems for the commercial
proteins.
Bio/technology
9 (1991) 1067-
1072.
r-factor Fabrc,
and processing is mediated
cerevisiae
precursor.
E., Nicaud,
of heterologous
proteins
solely by the pre-segment
in of
Lopez,
extracellular
M.C. and Gaillardin, and secretion
proteasc.
C.: Role of the
of the Yarrowiu
J. Biol. Chem.
lipo/ytica
266 (1991)
3782-
3790. Fabre,
E., Tharaud,
C. and Gaillardin,
is rescued
C.: Intracellular
by transcomplementation
transit of a yeast
with its prodomain.
P., Guyaneux,
L., Chasles,
M. and Gaillardin,
ur.7 sequence isolated in a morphogenesis
C.A.,
Gorman,
W.D.,
Wilhelm,
M.C. and
using fibrin glut in a rat mastec-
J. and
A., Sathe, G.M.,
Young,
N-glycosylated
interleukin-lp
in Sacchuromycrs
leader
from
ulbicans.
peptide
Cundida
P.R.:
Simon,
Secretion
J. Biol. Chem.
of
using
cerevisiae
a
266 (1991)
Matoba, S. and Ogrydziak, D.M.: A novel location for dipeptidylaminopeptidase processing sites in the alkaline extracellular protease of McDonagh, (Ed.),
J. Biol. Chem. 264 (1989) 6037-6043.
lipobkcc.
J.: Structure Hemostasis 2nd
and function
of factor
and Thrombosis.
ed. J.B.
XIIIa.
Basic
Lippincott
In: Caller,
Principles
Company.
B.S.
and Clinical
Philadelphia,
1987,
pp. 289-300. Muesch,
A., Hartmann,
E., Rohde,
oport, T.A.: A novel secretory
K., Rubartelli, pathway
Gaillardin,
C.
and
/&galactosidase
Ribet, activity
l@o/~fica. Curr.
Girolami,
Genet.
A., Sartori,
factor
C.: Scarcity
A., Sitia, R. and Rap-
for secretory
protease (XPRZ) gene of the yeast 12 (1989) 285-298.
mutant of the yeast Yarrowiu expression
resistance
of
Ycrrrowia
coding
of
G. and Ktipper,
for human
factor
H.A.: Char-
XIIIa.
Proc.
Natl.
Sci. USA 83 (1986) 8024-8028. H.: Genetics
Yurroti,iu
and
genetic
Adv.
lipolyfica.
Rinas,
engineering
Biochem.
produced
proteins?
TIBS
Eng.
of the industrial Biotcch.
yeast
43 (1990)
43-
L.E., Fujikawa,
K. and Davie, E.W.: Amino
Jagadeeswaran,
P. and Haas,
P.: Synthesis
of human
coagulation
25
factor
XIII in yeast. Gene 86 (1990) 279-283. Kreil, G.: Processing of molecular
of precursors
ticketing.
TIBS
by dipeptidylaminopeptidases: 15 (1990) 23-26.
Yurrowiu
R., Broker,
G.: Characterization
in Socchcrromyces cereksiue. F., Tosetto,
pharmacokinetics
C.:
cxtracellular J. Biotechnol.
lipolJGu.
M., Kargcs,
H.E.,
ofrccombinant
Kilpper,
factor XIIIa
Bio/Technology
A., Di, Bona,
8 (1990) 543-
E. and Castaman,
of a placenta-derived
factor
type I and type 11 factor XIII deficiency. 30-34. Sleep, D.. Belfield, G.P. albumin
and Goodey,
from the yeast
different leader sequences. Takahashi,
N., Takahashi,
blood coagulation
acid sequence of the a subunit of human factor XIII. Biochemistry (1986) 6900-6906.
P. and Gaillardin,
of the alkaline
546.
strum
73. A., Hendrickson,
and amplification
U., Risse, B., Jaenicke,
Rodeghiero, classification
77 (1991) 565-569.
E., Zcttlmeissl,
of cDNA
in
11 (1987) 369-375.
Br. J. Haematol.
U., Amann,
acterization
directed
M.T. and Simioni, P.: An updated
XIII defects.
Grundmann,
LELI2
phleomycin
J.M., Fournier,
of
sequencing
H.A. and Zettlmeisl,
A.M.: and
Nicaud, J.M., Fabre, E., Beckerich, Cloning,
Yeast 7 (1991) 25-36.
lipo!viicu.
Ichinose,
Meyers,
J.
Biol. Chem. 267 (1992) 15049-15055. Fournier,
Heslot,
Spotnitr,
prevention
15 (1990) 86-88.
protease
Acad.
P.L.,
Practice,
DNA 7 (1988) 355-360.
J.-M.,
pro region in the production alkaline
T.M.,
Seroma
Livi, G.P., Lillquist, J.S., Miles, L.M., Ferrara,
Yurrowiu
Ernst, J.F.: Efficient secretion Saccharomyces
Masterson,
R.F.:
15348-15355.
technol. 34 (1991) 756-764. Buckholz, R.G. and Gleeson, M.A.G.: production
W.H.,
Morgan,
tomy model. Arch. Surg. 125 (1990) 305-307.
13888-13889.
Broker, M., Bauml, O., Gilttig, A., Ochs, J., Bodenbenner, E.: Expression
Lindsey,
human Traore,
placenta.
Am. J. Hematol.
A.R.:
Bio;Technology
Proc. Natl. Acad. J.-C.:
of human using five
cerevisiue
8 (1990) 42-46. F.W.: Primary
factor XIIIa (fibrinoligase.
F. and Mcunicr,
in
36 (1991)
The secretion
SacchuromJres
Y. and Putnam,
G.: Clinical
XIII concentrate
structure
transglutaminase)
of
from
Sci. USA 83 (1986) 8019-8023.
Cross-linking
of caseins
by human
pla-
cental factor XIIIa. J. Agric. Food Chem. 39 (1991) 1892-1896. von Heijne, G.: A new’ method
for predicting
signal sequence
cleavage
sites. Nucleic Acids Res. 14 (1986) 4683-4690. a case
Yon, J. and Fried, M.: Precise gene fusion by PCR. Nucleic Acids Res. I7 (1989) 4895.