The Intracellular Production and Secretion of HIV-l Envelope Protein in The Methylotrophic Yeast Pichia Pastoris
The Intracellular Production and Secretion of HIV-l Envelope Protein in The Methylotrophic Yeast Pichia Pastoris
The Intracellular Production and Secretion of HIV-l Envelope Protein in The Methylotrophic Yeast Pichia Pastoris
0 1993 Elsevier Science Publishers B.V. All rights reserved. 0378-1119/93/$06.00 111
GENE 07555
Carol A. Scorera, Richard G. Buckholz b7* , Jeffrey J. Clarea and Michael A. Romanosa
‘Wellcome Research Laboratories, Beckenham, Kent BR3 3BS, UK; and bThe Salk Institute Biotechnology/Industrial Associates Inc. (SIBIA),
P.O. Box 85200, San Diego, CA 921 M-9268. USA
Received by G.P. Livi: 11 February 1993; Revised/Accepted 9 July/l 1 August 1993; Received at publishers: 29 August 1993
SUMMARY
The human immunodeficiency virus type 1 (HIV-l) envelope glycoprotein, gp120 (ENV), is required in large quantities
for immunological studies and as a potential vaccine component. We have expressed the DNA encoding gp120 in a
highly efficient expression system based on the methylotrophic yeast, Pichia pastoris. The native gene was found to
contain a sequence which resembled a Saccharomyces cerevisiae polyadenylation consensus and acted as a premature
polyadenylation site in P. pastoris, resulting in the production of truncated mRNA. As full-length mRNA was produced
in S. cerevisiae, this indicates differences in mRNA 3’-end formation between the two yeasts. Inactivation of this site by
site-directed mutagenesis revealed several additional fortuitous polyadenylation sites within the gene. We have designed
and constructed a 69%-synthetic gene with increased G+ C content which overcomes this transcriptional problem,
giving rise to full-length mRNA. High levels of intracellular, insoluble, unglycosylated ENV were produced [ 1.25 mg/ml
in high-density (2 x 10” cells per ml) fermentations]. ENV also was secreted from P. pastoris using the S. cerevisiae
a-factor prepro secretion leader and the S. cereuisiae invertase signal sequence. However, a high proportion of the
secreted product was found to be hyperglycosylated, in contrast to other foreign proteins secreted from P. pastoris.
There also was substantial proteolytic degradation, but this was minimized by maintaining a low pH on induction.
Insoluble, yeast-derived ENV proteins are being considered as vaccine antigens, and the P. pastoris system offers an
efficient method of production.
Correspondence to: Dr. CA. Scorer, Department of Cell Biology, gp160, 160-kDa envelope glycoprotein of HIV-1 that is cleaved into
Wellcome Research Laboratories, Langley Court, Beckenham, Kent gp120 (120-kDa) and gp41(41-kDa); HIM, gene encoding tri-functional
BR3 3BS, UK. Tel. (44-81) 658-2211; Fax (44-81) 650-0725. enzyme: phosphoribosyl-ATP cyclohydrase, phosphoribosyl-ATP pyro-
*Present address: Department of Molecular Biology, Glaxo Research phosphohydratase and histidinol dehydrogenase; HIV-l, human immu-
Institute, Five Moore Drive, Research Triangle Park, NC 27709, USA. nodeficiency virus type 1; kb, kilobase or 1000 bp; MCS, multiple
Tel. (1-919) 941-3308. cloning site (polylinker); MFal, gene encoding a-factor; nt, nucleo-
tide(s); oligo, oligodeoxyribonucleotide; P.. Pichia; Pl, sequence
Abbreviations: A, absorbance (1 cm); aa, amino acid(s); AOXl, gene (S-ATTATTTTATAAA) resembling known S. cereuisiae polyadenyla-
encoding alcohol oxidase; bp, base pair(s); CHO, Chinese hamster tion consensus; PA, polyacrylamide; PHOS, gene encoding acid phos-
ovary; E., Escherichia; ENV, polypeptide (54 kDa) or glycoprotein phatase; S., Saccharomyces; SDS. sodium dodecyl sulfate; SUC2, gene
(gp120) product of ENV; ENF, DNA encoding HIV-l,,, gp120; encoding invertase; UTR, untranslated region(s).
112
ceptor, CD4 (McDougal et al., 1986), and its presence on mutant strain, rnnn9, a product of about 120 kDa was
the surface of infected cells leads to the formation of syn- produced which had improved immunological properties
cytia (Sodroski et al., 1986). ENV is a major target for (Hitzeman et al., 1990). However, mnn9 strains exhibit
both antibody-mediated neutralization and cytotoxic im- slower growth and greater osmotic sensitivity than wild-
munity (Bolognesi, 1990) and as such is a likely candidate type cells and are unlikely to be suitable for large-scale
for inclusion in a future subunit vaccine against production.
HIV-l. The methylotrophic yeast Pichia pastoris has been de-
Fragments of gp160 have been produced intracellularly veloped as an efficient system for high-level production
in Escherichia coli (Putney et al., 1986) and of foreign proteins (Cregg et al., 1987; Sreekrishna et al.,
Saccharomyces cereuisiae (Barr et al., 1987); these pro- 1989; Clare et al., 1991a, b; Romanos et al., 1991a). A
ducts are unglycosylated and have non-native conforma- major advantage of this organism is the ease with which
it can be grown to extremely high cell density (130 mg
tions. Secreted, glycosylated HIV-l envelope proteins
dry weight per ml of culture) in a simple defined medium.
have been produced in a wide variety of systems, includ-
A number of proteins have been secreted to exceptionally
ing recombinant baculovirus (Wells and Compans, 1990),
high levels; for example: S. cerevisiae invertase, 3 mg/ml
adenovirus (Dewar et al., 1989), vaccinia virus (Barrett
(Tschopp et al., 1987); human serum albumin, 3 mg/ml
et al., 1989) and Chinese hamster ovary (CHO) cells
(K. Sreekrishna, personal communication); mouse epi-
(Berman et al., 1989). However, expression levels were
dermal growth factor, 0.45 mg/ml (Clare et al., 1991b).
low, making large-scale production difficult. These products may be in an almost pure form in the
The secretion of ENV from S. cerevisiae has been re- culture medium. Furthermore, most of the secreted inver-
ported, but this was shown to be in a ‘hyperglycosylated tase was not hyperglycosylated, and did not contain im-
form of greater than 600 kDa, which reacted poorly with munogenic terminal 1,3-linked mannose residues
antibodies to mammalian cell-derived material, possibly (Tschopp et al., 1987; Trimble et al., 1991) suggesting
due to masking of epitopes by carbohydrate (Hitzeman that P. pastoris may be more suitable for the production
et al., 1990). A similar problem had been previously ob- of foreign glycoproteins than S. cereuisiae. We therefore
served with Epstein-Barr virus membrane antigen, gp350, chose to investigate the synthesis of ENV in l? pastoris
secreted from yeast (Schultz et al., 1987). When ENV was and to determine its potential for large-scale ENV
produced in an S. cerevisiae hyperglycosylation-deficient production.
TABLE I
Summary of ENV expression vectors
(a) (b) 3’ end (data not shown). The truncated transcript was
found to be in the polyadenylated RNA fraction after
1 2 3 M 1 2 3 4
kb
oligo(dT)-cellulose chromatography (data not shown).
A number of consensus sequences have been implicated
in S. cerevisiae 3’-end formation, including the tripar-
tite sequence TAG..(T-rich)..TA(T)GT..(A +T-rich)..TTT
(Zaret and Sherman, 1984) and TTTTTATA (Henikoff
and Cohen, 1984). More recent evidence suggests that a
variety of different signals, acting unidirectionally and
bidirectionally, are used in yeast (Irniger et al., 1991).
I) Poorly-defined sequences downstream from the consen-
sus appear to be involved in determining both the site
‘
and the efficiency of 3’-end formation (Zaret and
Fig. 1. Northern blot analysis of mRNA from native, mutated and Sherman, 1984; Henikoff and Cohen, 1984). S. cerevisiae
partly synthetic ENV genes. P. pastoris GS115 (his4) was transformed
mRNA 3’-end formation, like that of higher eukaryotes,
with pPSV215-ENV, pPSV215-ENV* or pPIC9-ENV (see Table I)
using the sphaeroplast method (Cregg et al., 1985). Plasmids were di- involves cleavage and polyadenylation of a precursor
gested either with Sac1 or BglII. Total RNA was prepared from trans- mRNA, although in yeast these processes appear to be
formed cells induced with methanol (l%, v/v) for 8 h and analyzed by tightly coupled and occur within a much shorter distance,
Northern blotting as described previously (Romanos et al., 1991b).
near the 3’ end of the gene (Butler et al., 1990).
(Panel a) Lane 1:pPSV215-ENV; lane 2: pPSV215-ENV* (with mutated
Pl), showing low abundance multiple transcripts; lane 3: pPIC9-ENV Examination of the HIV-l gp120 coding region re-
(containing a partly synthetic ENI;; see section b); M: RNA markers. vealed a sequence, ATTATTTTATAAA (denoted Pl),
(Panel b) RNA from a pPSV215-ENV* transformant induced for 8 h close to the presumed 3’ end of the 700~nt transcript,
probed with synthetic oligos designed to hybridize to different regions
which resembled the TTTTTATA consensus sequence;
of the ENV mRNA, as indicated (position 1 is the mRNA start point).
Lane 1: 71-86; lane 2: 174209; lane 3: 699-729; lane 4: 7944824. The this sequence corresponds to nt 6303-63 15 in the HIV,*,
small arrow indicates the position of AOXI mRNA (2204 nt) which sequence published by Wain-Hobson et al. (1985). We
also hybridizes to the 5’ probe used for lane 1. The large arrow indicates altered Pl by site-directed mutagenesis (Zoller and
the position of the full-length ENV transcript. The transcript from
Smith, 1983) to TTTCTTCTACAAG, yielding plasmid
pPIC9-ENV (Panel a, lane 3, large arrow) is slightly longer than the
full length transcript from pPSV215-ENV* (Panel b, large arrow), since
pPSV215-ENV*. Northern blot analysis indicated that
the former contains the u-factor prepro region (269 nt) whereas the pPSV215-ENV* gave rise to a series of longer, low-
latter has the PH05 signal sequence (51 nt). abundance transcripts, including some full-length ENV
mRNAs (Fig. la,b). A number of oligo probes were con-
RESULTS AND DISCUSSION structed which were designed to hybridize at different
positions along the length of the ENV mRNA. We ob-
(a) Transcription of HIV-l ENI’ in P. pastoris served that the shorter transcripts did not hybridize to
Preliminary experiments were carried out using the probes furthest downstream (Fig. lb, lanes 2-4), imply-
P. pastoris expression vector, pPSV215-ENV (Table I). ing that the mRNAs constituted a 5’-nested set, i.e., all
This contained DNA encoding the S. cerevisiae acid phos- transcripts initiated in the AOXl promoter, and had 3’
phatase (PHO5) secretion signal fused to the coding se- ends which mapped at different sites in ENV. This sug-
quence for mature HIV-l LAv gp120 (denoted ENV), gested that Pl was indeed part of a major site causing
under the control of the P. pastoris alcohol oxidase premature mRNA 3’-end formation and that when this
(40X1) promoter. The host strain, GS115 (hid) was site was inactivated by mutagenesis in pPSV215-ENV*,
transformed with pPSV215-ENV, which had been di- several sites further downstream were revealed; the fact
gested at the unique Sac1 site in the 5’ AOXl DNA, in that several low abundance transcripts were seen (Fig. 1b)
order to promote integration by homologous recombina- suggests that these sites were only partially active. Our
tion upstream from the chromosomal AOXl gene. On data do not allow us to distinguish between the alterna-
induction of transformants with methanol, no ENV pro- tive possibilities that these sites caused premature tran-
duct was detectable by Western blotting. Northern blot scriptional termination or processing of a full-length
analysis indicated that the ENV mRNA was truncated precursor mRNA. However, we will henceforth refer to
(Fig. la), having an approximate size of 700 nt instead of them as fortuitous polyadenylation sites, since the trun-
the expected 1900 nt. The 700-nt transcript hybridized cated mRNAs were polyadenylated.
with a probe derived from the AOXl 5’ untranslated The discovery of truncated mRNA from HIV-l ENV
region, but failed to hybridize to 3’ probes, suggesting in P. pastoris was unexpected because full-length ENV
that it had initiated correctly, but was truncated at the mRNA was produced in S. cerevisiae (data not shown).
114
. . .
w aI
. . .
1 TXGCTAC T3AAAAGTIGTGa;TCACTGmACTA cGG&&x' To~:ToGApIGGAAG~CcACC~C~TATrmGTGcplrrp 'AMGsu&ATACAGAGG~ACATAATG~
SATEKLWVTVYYGVPVWKEATTTLFCASDAKAYDTEVHN" 40
YSiT
121 TGDGCCACACATGCCTGDTDTACCCACAGACCCCMCCCACmGAAG TAGTATl'3GTAAAXXGACAGAAM'l-mTAACATo TXAAAAA'lGACATGGTAGAACAG~GGA',T,TA
WATHACVPTDPNPQEVVLVNVTENFNMWKNDM"SQ,4HEDI 80
Id1
361 RG~~A~~T~C~~~~TAAAAR ~TPCTACUGT'XGGACA~A~CA
SGEMMMEKGEIKNCSFNISTSIROKVQKEYAFPYKLDlIp 160
NCOI
601 GCTArr-G'ICjTRACAI\CPJSIACTITCIULCGGT
AILKCNNKTFNGTGPCTNVSTVQCTHGIRPVVSTQLLLNO 240
BclI
721 TxlnxcrrJww\AGAAG ~~~~AG~T~IAC~~~~L~A~I~D~~AAGACACC
SLAEEEVVIRSANFTDNAKTIIVQLNQSVEINCTRPNNNT 280
lxa* XlnI
841 AG~AGAAvxAAAGA GGTCCAGGTAGAGC~CGITA~GGTAAGATT~~TAACA'P.?AGA~MGCTCAC'~GTAACATC~GCTAAG-CGCTACCTT~.AG
RKSIRIQRGPGRAFVTIGKIGNMRQAHCNISRAKWNATLK 320
BstEII Ecom
961 CAAATC iz%cAA -GA~~mGGTAACRRU\AGACCArrPlrrmAAG ~'~X&UCCAGAMTGTCAC'PZAT~CC~AAC~G~GG~~~
QIASKLREQFGNNKTIIFKQSSGGDPEIVTHSFNCGGEFF 360
YCNSTQLFNSTWFNSTWSTEGSNNTSGSDTITLPCRIKQF 400
0-1 Ml"1
1201 ArrAACATGTGGCARGAAGTT~~CCA~TACoCC BBTBTCACTGGTCTACTATPG UiB3!ZACGGTXTAPCAACAACAAC
INMWQEVGKA~~YAPPISGQIRCSSNITGLLLTKDGGNNNN 440
PflMI dvrI1
1321 CZTlrmAAATCT ~CA~AC~C~~A~C~~TACAPI
GSEIFRPGGGDHRDNWRSELYKYK"VKIEPLGVAPTKAKS 4*0
NOCI
1441 AGAGTPXCCR
RVVQREKR-
Fig. 2. Sequence of the partly synthetic HIV-l,,, ENV (gp120) gene. See section b for the considerations involved in the design of the synthetic
region,which extends from nt 1 to nt 43 (KpnI site) and from nt 455 (AM site) to the 3’ end. The intervening region (nt 44-454) is non-synthetic and
was obtained from the native HIV-l rAv ENV gene [nt 589446303; numbering according to Wain-Hobson et al. (1985)]. Underlined regions denote
restriction sites, as marked. Restriction sites in bold type are common to both the partly synthetic and native genes. The aa are aligned with the first
nt of each codon. The initiating ATG is derived from (i) the PHOS signal DNA (pSV215-ENV and pSV215-ENV*); (ii) the u-factor prepro region
DNA (pPIC19-ENV); or (iii) the SUC2 signal DNA (pPICll-ENV). In the intracellular expression vector, pPIC3-ENV, the initiating ATGimmedi-
ately preceded the first codon shown in the figure. aa numbering refers to the mature secreted protein.
For the major polyadenylation site, Pl, this difference rich (Zaret and Sherman, 1984; Henikoff and Cohen,
could have been due either to the fact that this differed 1984). We have observed fortuitous polyadenylation with
from the S. cereuisiae TTTTTATA consensus, or to a number of A+T-rich foreign genes expressed in
different requirements for sequences downstream from S. cerevisiae (unpublished results). Although high A + T
the consensus in the two yeasts. The presence of several content is not sufficient to cause premature polyadenyla-
additional polyadenylation sites active in P. pastoris, but tion, increasing G+C content by chemical synthesis has
not S. cerevisiae, indicates significant differences in the been shown previously to solve this problem for tetanus
signals for mRNA 3’-end formation between the two toxin fragment-C DNA (Romanos et al., 1991b). In con-
yeasts. trast to fragment-C DNA, ENV is only slightly A+T-
rich compared to yeast chromosomal DNA (S. cerevisiae
(b) Design and chemical synthesis of ENV and or l? pastoris), however, EN I/ contains A +T-rich runs
construction of expression vectors which may cause the problem in this case.
The poor conservation of known S. cerevisiae polyade- Three aspects were involved in the design of the syn-
nylation signals, their tolerance to significant sequence thetic region of ENV. Firstly, G+C content was
alterations (Osborne and Guarente, 1989), and the in- increased (from 37.6% to 43.6%) and special attention
ferred differences in l? pastoris meant that it was very was paid to eliminating A+T-rich runs. Secondly, the
difficult to predict the exact location of the several down- redundancy of the genetic code was used to insert unique
stream polyadenylation signals found in ENV DNA. restriction sites at regular intervals (approx. 100 bp). This
Therefore, since further site-directed mutagenesis was not would facilitate subsequent manipulations and allow re-
feasible, we decided to chemically synthesize the relevant placement of any regions containing synthesis errors.
part of ENV, from Pl to the 3’ end (1104 bp, or 69% of Thirdly, the synthetic region was designed to have an
the gene), increasing the G+C content. Known consen- optimal codon bias for high-level expression in yeast
sus sequences for 3’-end formation in yeast are all A + T- (Bennetzen and Hall, 1982) where this was in accordance
115
(b)
(a)
12345678 M 1 2 3 4 5 6 7. M
-46
Fig. 4. Intracellular expression of HIV-l ENV. Cultures were induced for 70 h in shake flasks and cell lysates prepared as described previously (Clare
et al., 1991a). The protein content of total cell extracts was determined using the Bio-Rad protein assay reagent. A 0.1% SDS-7.5% PA gel was used
to separate proteins (50 ug of protein per lane). Proteins were transferred to nitrocellulose using the Western blot technique. ENV protein was
visualized using a sheep polyclonal antiserum to CHO cell-derived gp120 (ADP401) and a second anti-sheep antibody conjugated to alkaline
phosphatase. (Panel a) Western blot of intracellularly expressed ENV from single and multi-copy pPIC3-ENV transformants induced in shake flasks.
Lane 1: 100 ng CHO cell-derived gp120 (ADP604) which also contains 50-kDa and 70-kDa breakdown products (this sample was not run on the
Coomassie blue stained gel shown in Panel b); lane 2: M4 (8 copy); lane 3: M3 (9 copy); lane 4: M2 (12 copy); lane 5: Ml (9 copy); lane 6: S2 (1
copy); lane 7: Sl (1 copy); lane 8: GS115 (untransformed host strain); M: protein markers. Single-copy transformants (lanes 6-7) showed very faint
immunoreactive bands which are barely visible in the figure. (Panel b) Coomassie blue stained gel of samples shown in Panel a. Lane 1: M4 (8 copy);
lane 2: M3 (9 copy); lane 3: M2 (12 copy); lane 4: Ml (9 copy); lane 5: S2 (1 copy); lane 6: Sl (1 copy); lane 7: GS115 (untransformed host strain);
M: protein markers. The arrow indicates the position of the ENV protein. The gel was scanned using a Joyce Loebl Chromoscan 3 to obtain an
estimate of the expression level of ENV.
expected size for non-glycosylated ENV (Fig. 4a, lanes in producing sufficient amounts of this material has led
2-5). In contrast, single-copy integrants gave very faint to the investigation of denatured ENV as a vaccine anti-
bands, indicating low-level expression (Fig. 4a, lanes gen. Studies in humans using yeast-derived insoluble
6-7). The four multi-copy integrants (Ml-M4) selected ENV fusions showed that this material is capable of prim-
gave similar yields of ENV as determined by scanning of ing CD4+ T cells specific for conserved regions of gp120
Coomassie blue stained SDS-PA gels (Fig. 4b). The copy (Abrignani et al., 1990).
number of integrated expression cassettes was determined
using quantitative DNA dot blots (Clare et al., 1991a): (d) Secretion of ENV
we estimated there to be 9, 12, 9 and 8 copies for Ml to Secretion of ENV in P pastoris was initially investi-
M4, respectively. The 12-copy transformant, M2, was se- gated using the a-factor secretion vector, pPIC9-ENV.
lected for induction in controlled fermentations, which Transformants were generated using either SacI- or
have been shown to be required for maximal yields in BglII-digested vector DNA. Digestion with BglII gener-
R pastoris (Clare et al., 1991a,b). In standard fermenter ates a fragment which can integrate by double-crossover
induction conditions M2 yielded ENV to 2.5% of cell recombination (transplacement) into the AOXl region,
protein (as judged by scanning SDS-PA gels) amounting and can be used to generate aoxl transformants. Single-
to 1.25 mg per ml of culture medium at 2 x lOlo cells crossover AOXl integrants also occur with BglII-di-
per ml. The product was entirely insoluble, as has been gested DNA due to in vivo ligation of the transforming
found for many secreted proteins when produced intra- fragment (Clare et al., 1991a). Single and multi-copy
cellularly. Despite the preference for native, glycosylated pPIC9-ENV transformants were distinguished by DNA
ENV containing conformational epitopes, the difficulty dot blot hybridization and induced cultures were ana-
117
lyzed in Western blots. All transformants secreted similar geneity of the products seen in Fig. 5 (lanes 3-8). Size
low levels of ENV in shake-flask inductions. The product heterogeneity is a common feature of heterologous glyco-
was detected as a heterogeneous smear (> 200 kDa to proteins produced in S. cereuisiae (Romanos et al., 1992).
~30 kDa), rather than a discrete band of approx. Recent evidence suggests that the nature of the secre-
120 kDa, suggesting that both hyperglycosylation and tion signal can dramatically affect the secretion kinetics
proteolytic degradation had occurred (data not shown). (Thill et al., 1990) and thus possibly the extent of glycosy-
Degradation was reduced by using medium buffered at lation of a protein in yeast. We constructed an alternative
pH 3.0 in shake flasks or at pH 5.0 in fermenters (see secretion vector (pPIC1 l-ENV), containing the
below), but surprisingly, not by the addition of bovine S. cerevisiae invertase (SUCZ) signal to investigate
serum albumin (at up to 1 mg/ml; data not shown). whether this could yield a product that was not hypergly-
A single-copy AOXl transformant, lA6, was induced cosylated. Unlike the a-factor prepro, the secretion signal
in a 2-litre fermenter. As with shake-flask inductions from invertase is a classical signal lacking a pro region.
secreted ENV was found to be hyperglycosylated, appear- Western blot analysis of the culture medium form
ing as a heterogeneous immunoreactive smear on pPIC1 l-ENV transformants showed that showed that
Western blots (Fig. 5). Deglycosylation revealed an ENV ENV was secreted at a low level and was hyperglycosy-
polypeptide of the expected size (approx. 59 kDa) as well lated to a similar extent as with the a-factor leader (data
as degradation products, although degradation was less not shown).
extensive than with unbuffered shake flask inductions. Both the pPIC9-ENV and pPICll-ENV products
The yield of secreted ENV was estimated to be between failed to react with a number of monoclonal antibodies
8 and 20 ug/ml, by comparison on a Western blot of the which had been raised to CHO cell-derived gp120, includ-
deglycosylated product with a CHO cell-derived gp120 ing one (ADP358) which reacts with the V3 loop. This
standard. Thus the two effects of hyperglycosylation and region comprises the principal neutralizing domain of
proteolytic degradation contribute to the extreme hetero- HIV-l (Javaherian et al., 1989). Lack of reactivity was
probably due to masking of epitopes by the extensive
glycosylation of the P past&s-derived product, as has
M 12 3456789 10 11 12 13 l4 15 16 17
been observed for an S. cerevisiae-derived hyperglycosy-
lated protein (Schultz et al., 1897). This conclusion is
supported by the fact that the deglycosylated
pPIC9-ENV product reacted more strongly with poly-
clonal antibodies raised to CHO cell-derived gp120
(ADP401) than the glycosylated forms (Fig. 5).
(e) Conclusions
(1) HIV-l ENF contains sequences which cause fortu-
itous polyadenylation in P. pastoris, but not in
S. cerevisiae. Mutagenesis of a sequence resembling the
TTTTTATA consensus for 3’-end formation in
Fig. 5. Secretion of HIV-l ENV. A single copy pPIC9-ENV trans-
S. cereuisiae mRNAs (Henikoff and Cohen, 1984) elimi-
formant, lA6, was grown to high cell density and induced with methanol
for 48 h in a Braun 2-litre fermenter (final A, of 420) as described by nated this polyadenylation site, but revealed the presence
Clare et al. (1991a). Samples were taken at various times after induction, of several additional sites further downstream. These re-
as indicated, for Western blot analysis. Cells were removed by centrifu- sults indicate significant differences in the signals for
gation and 48 ul of unconcentrated culture medium (or 25 pl of those
mRNA 3’-end formation in the two yeasts.
samples treated with endoglycosidase H) were applied per lane of a
0.1% SDS-7.5% PA gel. Proteins were transferred to nitrocellulose
(2) Chemical synthesis was used to increase the G+C
using the Western blot technique. ENV protein was visualized using content of the portion of ENI/ (the 3’ 69%) which con-
the polyclonal antiserum ADP401. M: protein markers; Lane 1: 100 ng tained the remaining fortuitous polyadenylation sites.
CHO cell-derived gp120 (ADP604); lane 2: 0 h; lane 3: 3 h; lane 4: 6 h;
This resulted in the elimination of the sites such that it
lane 5: 18 h; lane 6: 24 h: lane 7: 28 h; lane 8: 48 h; lane 9: control
fermenter induction of transformant containing a similar plasmid to
was possible to obtain full-length ENV mRNA in
pPIC9-ENV, but which lacked ENV, 50 h after induction. Lanes 10-17: P. pastoris.
as lanes 2-9, but using only 25 ul of supernatant treated for 18 h with (3) A 12-copy transformant yielded high levels
9 milliunits endoglycosidase H ( Boehringer-Mannheim) as described (1.25 mg/ml of culture) of intracellular, unglycosylated
previously (Romanos et al., 1991b). The heterogeneity of ENV in lanes
ENV (54 kDa) in fermenter inductions. As expected, the
3-8 is the result of a combination of hyperglycosylation and degrada-
tion. Lanes lo-16 in which the product is deglycosylated reveal the product was insoluble; however, similar products have
limited degradation which had occurred. been shown to elicit protective immune responses in man
118
(Abrignani et al., 1990). Owing to the difficulty in produc- Bolognesi, D.: Progress in vaccine development against SIV and HIV.
J. AIDS 3 (1990) 390-394.
ing large quantities of native gp120, such proteins are
Butler, J.S., Sadhale, P.P. and Platt, T.: RNA processing in vitro pro-
being considered as vaccine antigens and the P. pastoris duces mature 3’ ends of a variety of Saccharomyces cereuisiae
system offers an efficient method of production. mRNAs. Mol. Cell. Biol. 10 (1990) 2599-2605.
(4) ENV was secreted from P. pastoris into the culture Clare. J.J., Rayment, F.B., Ballantine, S.P., Sreekrishna, K. and
Romanos, M.A.: High-level expression of tetanus toxin fragment C
medium using the S. cerevisiae a-factor (MFaI) prepro
in Pichiu pastoris strains containing multiple tandem integrations
leader. Unexpectedly, a large proportion of the product of the gene. Bio/Technology 9 (199la) 455-460.
was hyperglycosylated, in contrast to other foreign pro- Clare. J.J., Romanos, M.A., Rayment, F.B., Rowedder, J.E., Smith, M.A.,
teins secreted from P. pastoris (Tschopp et al., 1987; Payne, M.M., Sreekrishna, K. and Henwood, C.A.: Production of
Grinna and Tschopp, 1989; J.J.C., unpublished). The mouse epidermal growth factor in yeast: high-level secretion using
Pichia pastoris strains containing multiple gene copies. Gene 105
secreted product showed substantial proteolytic degrada-
(1991b) 205-212.
tion, but this was minimized by maintaining a low pH Cregg, J.M., Barringer, K.J., Hessler, A.Y. and Madden, K.R.: Pichia
on induction. The combined effects of hyperglycosylation pastoris as a host system for transformations. Mol. Cell. Biol. 5
and proteolysis caused the secreted glycoprotein to show (1985) 3376-3385.
Cregg, J.M., Tschopp, J.F., Stillman, C., Siegal, R., Akong, M., Craig,
extreme size heterogeneity.
W.S., Buckholz, R.G., Madden, K.R., Kellaris, P.A., Davies, G.R.,
Smiley, B.L., Cruze, J., Torregrossa, R., Velicelebi, G. and Thill, G.P.:
High-level expression and efficient assembly of hepatitis B surface
Osborne, B.I. and Guarente, L.: Mutational analysis of a yeast tran- Thill, G.P., Davis, G.R., Stillman, C., Holtz, G., Brierley, R., Engel, M.,
scriptional terminator. Proc. Natl. Acad. Sci. USA 86 (1989) Buckholz, R., Kinney, J., Provow, S., Vedick, T. and Siegel, R.S.:
4097-4101. Positive and negative effects of multi-copy integrated expression
Romanos, M.A., Clare, J.J., Beesley, K.M., Rayment, F.B., Ballantine, vectors on protein expression in Pichia pastoris. In: Heslot, H.,
S.P., Makoff, A.J., Dougan, G., Fairweather, N.F. and Charles, I.G.: Davies, J., Florent, J., Bobichon, L., Durand, G. and Penasse. L.
Recombinant Borderella pertussis pertactin (P69) from the yeast (Eds.) Proceedings of the 6th International Symposium on the
Pichia pastoris: high-level production and immunological properties. Genetics of Microorganisms, Vol. II. Societe Francaise de
Vaccine 9 (1991a) 901-906. Microbiologic, Paris. France, 1990, pp. 477-490.
Romanos, M.A., Makoff, A.J., Fairweather, N.F., Beesley, K.M., Slater, Tschopp, J.F., Sverlow, G., Kosson, R., Craig, W. and Grinna, L.: High-
D.E., Rayment, F.B., Payne, M.M. and Clare, J.J.: Expression of level secretion of glycosylated invertase in the methylotrophic yeast,
tetanus toxin fragment C in yeast: gene synthesis is required to Pichia pastoris. Bio/Technology 5 ( 1987) 1305-1308.
eliminate fortuitous polyadenylation sites in AT-rich DNA. Nucleic Trimble, R.B., Atkinson, P.H., Tschopp. J.F., Townsend, R.R. and
Acids Res. 19 (1991b) 1461-1467. Maley, F.: Structure of oligosaccharides on Saccharomyces WC2
Romanos, M.A., Scorer, C.A and Clare, J.J.: Foreign gene expression invertase secreted by the methylotrophic yeast Pichia pastoris.
in yeast: a review. Yeast 8 (1992) 423-488. J. Biol. Chem. 266 (1991) 22807-22817.
Schultz, L.D., Tanner, J. Hofmann, K.J., Emini, E.A., Condra, J.H., Wain-Hobson, S., Sonigo, P., Danos, O., Cole, S. and Alizon, M.:
Jones, R.E., Kieff, E. and Ellis, R.W.: Expression and secretion in Nucleotide sequence of the AIDS virus, LAV. Cell 40 (1985) 9-17.
yeast of a 400-kDa envelope glycoprotein derived from Epstein-Barr Wells, D.E. and Compans, R.W.: Expression and characterization of a
virus. Gene 54 (1987) 1133123. functional human immunodeficiency virus envelope glycoprotein in
Sodroski, J., Goh, W.C., Rosen, C., Campbell, K. and Haseltine, W.A.: insect cells. Virology 176 (1990) 5755586.
Role of the HTLV-III/LAV envelope in syncitium formation and Zaret, K.S. and Sherman, F.: Mutationally altered 3’ ends of yeast
cytopathicity. Science 322 (1986) 470-474. CyCl mRNA affect transcript stability and translational efficiency.
Sreekrishna, K., Nelles, L., Potenz, R., Cruze, J., Mazzaferro, P., J. Mol. Biol. 177 (1984) 107-136.
Fish, W., Motohiro, F., Holden, K., Phelps, D., Wood, P. and Zoller, M.J. and Smith, M.: Oligonucleotide-directed mutagenesis of
Parker, K.: High-level expression, purification, and characterization DNA fragments cloned into Ml3 vectors. Methods Enzymol. 100
of recombinant human tumor necrosis factor synthesized in the (1983) 468-500.
methylotrophic yeast Pi&u pastoris. Biochemistry 28 (1989)
4117-4125.