without a cleavable signal peptide. We have tested whether the signal responsible for its membrane insertion is present within its transmembrane peptide using ...
The EMBO Journal vol.5 no.7 pp. 1543-1550, 1986
The transmembrane segment of the human transferrin receptor functions as a signal peptide
Marino Zerial, Paul Melancon, Claudio Schneider and Henrik Garoff European Molecular Biology Laboratory, Postfach 10.2209, 6900 Heidelberg, FRG Communicated by L.Philipson
The human transferrin receptor (TR) is a protein comprising 760 amino acid residues that spans the membrane once with its N terminus towards the cytoplasm. It is synthesized without a cleavable signal peptide. We have tested whether the signal responsible for its membrane insertion is present within its transmembrane peptide using a combined recombinant DNA/in vitro translation approach. The complete TR coding region was first reconstructed from overlapping TR cDNA clones and then engineered into an SP6-based transcription vector. In vitro transcription and subsequent translation in the presence of rough microsomes yielded TR molecules that were glycosylated and correctly inserted into the membrane. Two kinds of experiments demonstrated that the spanning region of the TR polypeptide contained the signal for translocation across the membrane of the rough endoplasmic reticulum. First, we deleted the spanning region of TR and showed that this deletion mutant could not be inserted. Second, we showed that two cytoplasmic proteins (the mouse dihydrofolate reductase and the chimpanzee a-globin) could be inserted into the microsomal membrane in the expected orientation when the TR transmembrane segment was added to their N termini. Thus, the spanning peptide was shown to be both necessary and sufficient for chain translocation. Further analyses demonstrated that the translocation event was dependent on the signal recognition particle. Key words: transferrin receptor/signal recognition/transmembrane segment
Introduction Plasma membrane (PM) proteins can be divided into three groups (I-IEl) on the basis of their orientation in the membrane (Wickner and Lodish, 1985; Garoff, 1985). Group I proteins span the lipid bilayer once and have their C terminus on the cytoplasmic side of the membrane. Group HI proteins have the opposite orientation in the membrane. Group III proteins span the membrane several times. The assembly of these various polypeptide chains in the rough endoplasmic reticulum (RER) membrane during synthesis has been explained by extending the signal hypothesis (Blobel and Dobberstein, 1975) to include also (i) internal signal peptides (or translocation signals) and (ii) stop transfer signals (Blobel, 1980; Sabatini et al., 1982; Walter et al., 1984; Wickner and Lodish, 1985; Garoff, 1985). For example, the translocation of group I proteins (e.g. class I histocompatibility antigens), which in general possess a cleavable N-terminal signal peptide, is thought to be initiated as for secretory proteins but is later interrupted by a second peptide located towards the C terminus. © IRL Press Limited, Oxford, England
This stop transfer signal is postulated to be part of, or equivalent to, the membrane-binding region of the protein. Group II proteins (e.g. human transferrin receptor; Schneider et al., 1983a) do not have a cleavable signal sequence. These proteins are postulated to contain an internal signal sequence within their transmembrane segment. This region possesses the hydrophobic characteristics of a signal peptide. The transmembrane segment of group II proteins is thought to insert into the RER membrane in such a way that the flanking N-terminal part of the chain is left on the cytoplasmic side. As synthesis continues the distal C-terminal part is transferred across the membrane to generate the final topology. Proteins which span the membrane several times (group Ill proteins, e.g. bovine opsin; Ovchinnikov et al., 1982; Hargrave et al., 1983; Friedlander and Blobel, 1985) could possibly be generated by the sequential expression of the various types of peptide signals already mentioned for group I and II proteins. In the present study, we have used recombinant DNA techniques and in vitro transcription and translation to test whether the transmembrane region of a group II protein functions as a signal peptide. As our model protein we have chosen the human transferrin receptor (TR). The primary structure of this molecule was recently deduced from the nucleotide sequence of cloned cDNA (Schneider et al., 1984). The TR molecule has a C-terminal extracellular domain of 671 amino acids and an N-terminal cytoplasmic one of 61 residues. The transmembrane region consists of a 28 amino acid residues long uncharged and hydrophobic peptide. Two kinds of experiments demonstrate that the TR indeed possesses a translocation signal (signal peptide) within its transmembrane region. First, we have deleted the spanning region of the TR and show that this deletion mutant cannot be translocated across RER membranes. Second, we show that two cytoplasmic proteins can be inserted into RER membranes when the TR transmembrane segment is fused to their N termini. Further analyses showed that the TR signal peptide requires signal recognition particle (SRP) for its function.
Results Translations of in vitro transcribed mRNA produce authentic TR The human TR mRNA has been cloned as a set of overlapping cDNA molecules (Schneider et al., 1984). Therefore, we first had to construct plasmid pTR which contains the complete coding region of the TR (see Figure la). For transcription, the complete TR cDNA was subcloned into the in vitro transcription vector pGEM 1 to yield plasmid pGEM 1 TR (Figure lb). Using SP6 polymerase, and by including a cap structure into the transcription mixture, we were able to synthesize microgram quantitites of capped TR mRNA from pGEM 1 TR. In vitro translation of this mRNA in the absence of rough microsomes yielded one major protein that migrates on SDS-polyacrylamide gel with an apparent mol. wt of 78 kd (Figure 2, lane 4). This product comigrates with the unglycosylated control TR which has been immunoprecipitated from tunicamycin-treated HeLa cells (lanes 2 1543
M.Zerial et al.
Acc
Acc
AccI
r pTR 43
(2)
Acc I
pTR 36
AccI Acc
Ac
oEco R
b
pTR I
Acc
,
(POEM)
(1) uXba
4'2.%
Ec)'7Ba H 1c
BarnHI
C
Pvu I Barn H
pGEM 1 TR g II P vu
Hae III
=- Bl3/ =pGEM 1 TR
Bam H /
TR1
coRI BmI / Barn HcDd
d
ie
Eo YDde
d
R
T
PvuI B -H
in
EcoRI
pGEM
pGEM 1 TR1 \~~~~~~~~~~~
E pGEM 2
TR -dhfr
Hae
SphBIH
2 I-ITpEM TRDdehfr
/
Sat
f*/PGE 2
g pGEM2 \
wTR A2-dhfr) /Hae IID
Alu l/
GEM
SphM
Sst I-
Sph
Fig. 1. Schemes of plasmids and their construction. (a) Reconstruction of the complete TR-coding sequence. Plasmids pTR36, 43 and 19 each contain overlapping TR cDNA sequences (in pEMBL 8) which in this order cover the complete TR-coding region from the 5' to the 3' end. The two unique AccI sites [marked AccI(2) and (3)] in the TR cDNA and the third one in the polylinker region of the vector molecule [AccI(1)], were used for the preparation of three fragments (shown by bolder lines in the circular plsamids). These were ligated into pTR which contains the complete TR-coding sequence. Circles represent DNA, the thicker parts of the circles indicate regions encoding mRNA sequences and the boxes delimit protein-coding regions. Recognition sites for relevant restriction endonuclesaes are indicated. The wavy lines border the region coding for the transmembrane peptide of TR. (b) Plasmids for in vitro transcription of TR mRNA , DHFR nmRNA and ca-globin mRNA. Explanations of plasmid circles are as a. The thick arrow indicates the SP6 promotor and the direction of transcription. (c) Construction of pGEM 1 TRA-1. Plasmid pGEM 1 TR was linearized with BamHI and then treated with Ba3l3 until the DNA region encoding the transmembrane peptide of the TR was removed. The Bal3 1-treated material was ligated to ClaI linker molecules and then restricted with ClaI and PvuI. The large fragments were isolated and ligated with the PvuI -BamHI and the BamHI -HaeIII/ClaI fragments as shown in the figure to generate pGEM 1 TR (/v) plasmids. In these constructions the very 5' end of the TR cDNA, which encodes the cytoplasmic NH2-terminal domain, has been recreated by including the BamHI -HaellI fragment in the ligation. The plasmid pGEM 1 TRA-l which was used in the in vitro translocation analyses, has been characterized by nucleotide sequencing and shown to be devoid of the DNA region encoding the transmembrane peptide. (d) Construction of pGEM 2 TR-dhfr and TR-globin. An EcoRI - DdeI fragment from pGEM 1 TR containing the membrane-spanning segment coding region was made blunt and then ligated wih EcoRI/BamHI- and EcoRI/NcoI-restricted (and end-polished) plasmids pGEM 2-dhfr and pGEM 2-globin, respectively. Explanation of plasmid circles are as in a. (e) Construction of pGEM 2 TRA 2-dhfr. This plasmid contains a deletion within the TR DNA encoding for the cytoplasmic tail. The deletion is located betwen the AluI site (position 275 in the TR map) and the HaeIlH site (422) of the TR sequence. The plasmid was obtained by ligating three fragments (SphI-SstI, SphI-AluI, HaeIII-SstI) isolated from plasmid pGEM 2 TR-dhfr.
1544
Biosynthesis of the transferrin receptors
5). Most likely the 85-kd protein corresponds to the core glyC= 1-
cosylated TR that was also observed by Schneider et al. (1983a,
pOEM 1 c c= TR e °=
I-
TR TM
2
,_
1983b, 1984) upon membrane-supplemented in vitro translations of HeLa and Molt-4 cell mRNA. The TR molecule is known to contain three asparagine-linked carbohydrate units (see references above). In lane 6 the two weaker bands between the 85- and 78-kd products most likely represent TR molecules with only one or two oligosaccharide chains, respectively. Protease digestion of the membrane-supplemented translation mixture resulted in one major protected polypeptide of 78 kd and minor bands of 76 kd and smaller. This is very similar to the effect of protease treatment on TR molecules which have been inserted into microsomes in vivo in HeLa cells and in vitro using HeLa cell mRNA (Schneider et al., 1983a). The result is explained by the preferential digestion of the N-terminal protein domain exposed to the cytoplasm. This means that the TR polypeptides we have synthesized in vitro are correctly inserted in the added microsomal membranes. The 78- and 76-kd forms probably correspond to the protected portions of the completely and partially glycosylated TR molecules discussed above. Note that the band of -67 kd seen in all lanes of in vitro synthesized TR is unspecific as it also is present in the translations without TR mRNA (lanes 9 and 10). The bands of low mol. wt seen in all TR translations might represent prematurely terminated products or possibly result from internal initiation of translation. The transmembrane peptide of TR is necessary for its insertion into the RER membrane The possibility that the transmembrane region of the TR polypeptide contains the translocation signal was investigated in two ways. First, we constructed a transmembrane peptide-deletion mutant of TR in order to see whether the spanning region was obligatory for chain translocation. Second, we fused the transmembrane region of the TR to cytoplasmic proteins to find out whether the hydrophobic TR peptide was sufficient for the generation of group II polypeptide topology. The deletion mutant pGEM 1 TRAI was generated from pGEM 1 TR using Bal31 exonuclease digestion (Figure lc). As shown schematically in Figure 3 the predicted mutant protein has 37 residues deleted. These are the 28 residues in the uncharged peptide segment plus seven and two flanking residues on its N- and C-terminal side, respectively. Thus, both hydrophilic protein domains are left almost intact by the deletion. Translation of the pGEM 1 TRAI mRNA in the absence of membranes yielded a protein with an apparent mol. wt slightly smaller than the wild-type product, as expected (compare lane 2 with lanes 1 and 3 in Figure 4). When translation was carried out in the presence of membranes the TRA I mutant migrated with the same mobility as when synthesized in their absence (compare lane 8 with lanes 2 and 7). There was no shift into a higher mol. wt form indicative of glycosylation, in contrast to what was observed for the wild-type TR (lane 4). This suggests that the TRAI protein has not been glycosylated and hence not inserted into the added membranes. Such an interpretation is supported by the absence of protected material after protease treatment (lane 9). We conclude that the mutant protein cannot be inserted into RER membranes. Evidently the presence of the transmembrane peptide of TR is necessary for the translocation process. The transmembrane region of TR is sufficient to translocate dihydrofolate reductase and a-globin across the RER membrane To test whether the transmembrane peptide of TR contains the information necessary to insert a polypeptide into the RER membrane as a group II protein, we constructed cDNA fusions design-
97
69
m
I
46
La I
._
MMM PP TX
M
1 2 3 4 5 6 7 8 9 10 11 Fig. 2. Fluorogram of SDS -PAGE analyses of [35S]methionine-labelled proteins obtained by translating pGEM 1 TR mRNA in a rabbit reticulocyte lysate in the absence (lane 4) or in the presence (lanes 6-8) of microsomal membranes. The membrane-supplemented sample was divided into three parts: one part was analysed directly (lane 6), another was treated with protease before analyses (lane 7) and the third with protease and Triton X-100 (lane 8). Control TR was run in lanes 2, 3 and 5. This has been immunoprecipitated from HeLa cells labelled with [35S]methionine in the presence (lanes 2 and 3) and absence of tunicamycin (lane 5). The mol. wt markers (lanes 1 and 11) are phosphorylase B (97 kd), albumin (69 kd), ovalbumin (46 kd), carboxyanhydrase (30 kd), lactoglobulin A (18 kd). Lanes 9 and 10 represent analyses of translation reactions (-+ membranes) in which the TR mRNA has been omitted.
and 3). We conclude that our in vitro product represents fullsized TR polypeptide. Translation in the presence of membranes (lane 6) yields an additional major product of 85 kd. The latter protein migrates on the SDS gel somewhat faster than the mature and fully glycosylated TR (90 kd) isolated from HeLa cells (lane
1545
M.Zerial et al. Transcription plasmid
Translated protein
pGEM 1-TRF
Translocation
I.
pGEM 1 -TRAE
L
+
pOEM
=
r
pGEM 2-globin pGrM 2-TR
pGEM
pGEM 1 TR A I
144.
gobin
3.9
r~~~
I
W
I TR
A
3
~A
I
I
i~
2-dh7r
97 96 ,|
pGEM 2-TR dhfr
pGEM 2-TR A2 dfr
=
+
I. . 4 q.
to
m. *.F
Fig. 3. Schi}emie of TR mLIutaInts
and
hybrids
and t heil
COMpetence
69
_
fiwr being.
translOCclted.
ed to express the spanning peptide of TR in front of either a-globin or dihydrofolate reductase (DHFR). The hybrid constructions were done by joining (i) the cDNA region encoding the cytoplasmic and the transmembrane domains of TR to the coding sequences of mouse DHFR (plasmid pGEM 2 TR-dhfr) (Figure Id), (ii) the same TR sequences to the coding sequences of chimpanzee at-globin (plasmid pGEM 2 TR-globin) (Figure Id) and (iii) the transmembrane region, lacking most of the cytoplasmic domain sequences, to the dhfr cDNA (plasmid pGEM 2 TRA2-dhfr, Figure le). The fusion proteins predicted from plasmids pGEM 2 TR-dhfr and pGEM 2 TR-globin contain a total of 96 N-terminal amino acid residues of TR in front of the complete DHFR and globin sequences, respectively (Figure 3). The TR DHFR fusion protein had two additional amino acid residues (Gly-Ala) at the fusion region as a consequence of the genetic engineering. In contrast, the TR a-globin construct contains no extra residues. The hybrid predicted to be expressed from pGEM 2 TRA2-dhfr contains a 43-residue fragment encompassing the TR-spanning region together with three residues from the TR N terminus in front of the DHFR protein (see Figure 3). In vitro translation of mRNA transcribed from the three plasmid constructs yielded protein products which had the expected mobility in SDS gels (see Figure 5, lanes 6 and 7, and 11 and 12 and Figure 6, lanes 6 and 7). The same pattern was seen both in the presence or absence of rough microsomes. This is in agreement with the fact that no potential glycosylation sites are present in any of the hybrid proteins. However, when membrane translations were treated with protease, large protected fragments were generated from all hybrid protein samples (Figure 5, lanes 8 and 13 and Figure 6, lane 8). No such fragments were seen in similar analyses involving the intact reporters (Figure 5, lane 4 and Figure 6, lane 3). The fact that the protease-resistant portion of the TR reporter hybrid appeared in each case to be somewhat larger than the corresponding reporter molecule alone (compare lanes 8 and 13 with 2 in Figure 5 and lane 8 with 1 in Figure 6) strongly suggests that the hybrids have been inserted across the membranes of the microsomes such that the C-terminal reporter portion is in the lumen and the N-terminal part, which is derived from the cytoplasmic TR domain, is exposed on the external surface. The reporter plus the spanning segment will thus correspond to the protease-resistant fragment of the hybrid. 1546
6M sA~ *W..%*.
46
4w4
~
*
D
-
I~~~~0
-
-
...
M M MI p p
T)
1
2
3 4
.x
M
M
M
p
p
IX
5 61 7 8 9 10 11
Fig. 4. Fluorogram of SDS PAGE analyses of [35S]methionine-labelled proteins obtained by translating pGEM 1 TRA I mRNA and pGEM 1 TR (as control) in a rabbit reticulocyte lysate. Addition of membranes (M), protease (P) and Triton X-100 (TX) is indicated. -
We conclude that all the information which is required for the insertion of the TR across ER membranes is confined within its N-terminal portion. Furthermore, the ability of TRA2-DHFR to be inserted clearly demonstrate that the transmembrane region itself contains the signal sequence. The translocation signal requires SRP for its jfinction The requirement of the TR signal peptide for SRP was studied in a translation assay composed of wheat germ lysate and saltwashed membranes. Using this system, which is essentially devoid of endogenous SRP activity, it has been shown that secretory proteins and many membrane proteins require the addition of exogenous SRP in order to be translocated (Hortsch
Biosynthesis of the transferrin receptors
pGEM 2TR dhfr
2pGEM 3g: ZEdh dhfr
pGEM 2TR A 2 dhfr
69 46 30
18
M M P
M P
TX
1
2
3
4
5
M MM PP P P TX TX 7 8 9 1011 12 13 14
MM M
6
Fig. 5. Fluorogram of SDS -PAGE analyses of [35S]methionine-labelled proteins obtained by translating plasmids pGEM 2-dhfr, pGEM 2 TR-dhfr and pGEM 2 TRA2-dhfr in a rabbit reticulocyte lysate. Addition of membranes (M), protease (P) and Triton-100 (TX) is indicated.
and Meyer, 1984). Furthermore, SRP can result in an inhibition of translation or an arrest of chain elongation when added without accompanying salt-washed membranes. Figure 7, lanes 4, 5, 13 and 14 shows the effect of SRP on the translation of mRNA transcribed from pGEM 2 TRA2-dhfr and pGEM 2 TR-globin in a wheat germ system without membranes. SRP inhibited the synthesis of the two hybrid proteins by 90% and 87%, respectively. In contrast, almost no inhibition was observed in the translations programmed with mRNA for the intact reporter molecules (lanes 2, 3, 11 and 12). Analysis made in the presence of membranes (lanes 6-9 and 15- 19) show that both TRA2-DHFR and TR-globin translocated 3.3 and 9 times more efficiently in the presence of exogenous SRP than without. The low amount of translocation activity which is seen in the absence of added SRP is probably due to some remaining SRP in the washed membranes. We conclude that the TR translocation signal requires SRP to function.
Discussion The current hypothesis for the generation of group II proteins is that these proteins contain an internal signal sequence within the transmembrane segment. We have shown here that this is the case for one typical group II protein, the human TR molecule. More specifically, the TR translocation signal was localized to a region encompassing amino acid residues 55-96. This peptide contains as its major part the uncharged and hydrophobic membrane-spanning segment of the TR molecule (residues
62-89). The most striking common feature of all characterized signal sequences is a stretch of at least seven hydrophobic and uncharged amino acid residues (Watson, 1984; von Heijne, 1985). Therefore, it is reasonable to assume that the translocation signal of the TR molecule is confined to this (spanning) segment and not to the short hydrophilic flanking regions which were left in our hybrid constructions. We have shown furthermore that the translocation signal of TR needs SRP to function in vitro. Our current model for TR topogenesis in the RER is schematically depicted in Figure 8. The translation of the N-terminal, cytoplasmic protein domain occurs on free ribosomes. The synthesis of the membrane-spanning segment induces the binding of SRP on the translating ribosome and the subsequent do-cking of the complex on the RER membrane. Chain translocation is then throught to be initiated through the looping of the transmembrane region into the lipid bilayer in such a way that the N-terminal domain is left on the outside and the distal chain is brought across the membrane. Recently an approach similar to ours has been used to show that the mouse liver asialoglycoprotein receptor (another group II protein) also contains an internal signal sequence within its spanning peptide segment (Spiess and Lodish, 1986). Using an in vivo expression approach, Bos et al. (1984) and Markoff et al. (1984) showed that the translocation of influenza neuraminidase is mediated by its N-terminal hydrophobic segment. Therefore, the proposed mechanism of biosynthesis appears to be general for this group of cell surface (and viral) proteins. At present we cannot tell exactly where in the 28-residue un1547
M.Zerial et al.
pOEM gloin
et al., 1986). However, the actual sequence of the hydrophobic residues appears unimportant. In Escherichia coli artificial sequences of hydrophobic residues function as well as natural ones (Davis and Model, 1985). On the other hand, typical cleavable signal peptides display three characteristic regions: (i) a hydrophilic (often basic) N terminus; (ii) a central hydrophobic core of at least seven residues; and (iii) a particular amino acid distribution at the cleavage site (Watson, 1984; von Heijne,
pOEM 2 TR globin
-
F 46
30
_-
mo15 M M M
M
TX 2
3e
4
8
M M p p
p p
1
1
TX
5
6
7
Fig. 6. Fluorogram of SDS PAGE analyses of [35S]methionine-labelled proteins obtained by translating plasmids pGEM 2-globin and pGEM 2 TRglobin in a wheat germ lysate. Addition of membranes (M), protease (P) and Triton-100 (TX) is indicated. -
charged peptide of the TR molecule the two topogenic sequences (membrane anchor and signal sequence) are located. In particular, we would like to know whether these are overlapping, either partially or completely, or whether separate peptide units are involved. Eventually, we would like to define in detail those structural features of the signals that are important for their function. The properties of transmembrane polypeptide segments have been extensively studied using group I proteins (see references below). In this case that region is well separated from the signal sequence located at the N terminus. The results showed that both the overall hydrophobicity of the transmembrane region and the basic residue(s), which are always flanking the latter region on its cytoplasmic side, are both important features for a stable membrane association (Davis et al., 1985; Davis and Model, 1985; Adams and Rose, 1985a, 1985b; Cutler and Garoff, 1986; Cutler
1548
1985). If we use the structural criteria mentioned above for the stop translocation and the signal peptide to define the two topogenic signals of the TR, one could tentatively assign both of these to the N-terminal portion of the membrane-binding polypeptide region. The flanking N-terminal basic residues [Lys (58) Lys (60 Arg (61)] would thus represent region (i) of the signal as well as the basic cluster of the anchor. The critical hydrophobic region of the TR signal sequence would be located in the adjacent spanning segment of the TR. The consensus features for signal peptidase recognition and cleavage would of course be absent in this internal signal peptide. The exact localization of the TR translocation signal could be tested by studying the in vitro translocation of (a) mutants with partial deletion of their transmembrane segment and (b) fusion proteins containing various parts of the same
segment. Materials and methods Materials All reactions were carried out in Eppendorf tubes sterilized by autoclaving. Restriction endonucleases, E. coli DNA polymerase I (Klenow fragment), T4 DNA polymersae and SP6 polymerase were obtained from Boehringer (Mannheim). Mung bean nuclease (FPLC grade) was obtained from Pharmacia (Freiburg), Bal31 nuclease from Bethesda Research Laboratories (Rockville), RNasin from Promega Biotech (Madison). T4 DNA ligase was kindly provided by F.Winkler (EMBL). The DNA linkers (ClaI 8-, 10-, 12-mer, EcoRI 8-mer) and the SP6 promoter sequencing primer were obtained from Boehringer (Mannheim). Ribonucleotides (from equine muscle) and spermidine used in the in vitro transcription were from Sigma (St Louis). The cap analogue m7G(5')ppp(5')G, the ultrapure deoxy- and dideoxy-ribonucleotides used for DNA sequencing were from Pharmacia. Ribonculease A, bovine serum albumin (BSA, Fraction V),
tunicamycin, proteinase K, phenylmethylsulfonyl fluoride (PMSF), diethylpyrocarbonate (DEPC), creatine phosphate and creatine phosphokinase were from Sigma. Rabbit reticulocyte lysate and stabilized L-[35S]methionine (S.J. 1515) were purchased from'Amersham (Braunschweig). Enhance was from New England Nuclear (Dreieich). Sephadex G-50, Sephacryl S-300 and protein A-Sepharose CL-4B were obtained from Pharmacia. LMP agarose, ultra pure acrylamide and N,N'-methylenebisacrylamide were from Bethesda Research Laboratories. Wheat germ lysate was prepared according to Grossman et al. (1980). Salt-washed dog pancreas rough microsomes (50 A280/ml) were kindly provided by B.Dobberstein (EMBL). SRP isolated from dog pancreas rough microsomes according to Walter and Blobel (1980) was a generous gift of D.Meyer (EMBL). DNA engineering Nucleic acid methods were as described in Maniatis et al. (1982), Dillon et al. (1985) and as recommended by the manufacturers of the various modifying enzymes used. The plasmid constructions were carried out as follows. PlasmidpTR (codes for the complete TR molecule). This was done using the pTR plasmids 36, 43 and 19 as shown in Figure la. Our DNA engineering was facilitated by three suitably located AccI sites, which, upon cleavage, each produced a unique sticky end [(1) CGAC, (2) AGAC, and (3)ATAC]. The 5' region of the TR cDNA was contained within the AccI (1) and (2) sites and this DNA fragment could be isolated from pTR36. The middle region which was present between the AccI (2) and (3) sites was isolated from pTR43. The rest of the TR cDNA including the vector part was obtained as a large AccI fragment from pTR19. Ligation of all three fragments yielded the plasmid pTR which contains the complete 2280-bp coding sequence of the TR cDNA together with most of its 5'and 3'-non-coding regions. Its identity was confirmed by extensive restriction endonuclease mapping, including analyses by AccI. Plasmid pGEM 7R (the 77 sequence in the transcription plasmid). To obtain pGEM 1 TR (Figure lb), the TR coding region was excised from pTR as a 2366-bp XhoI-XbaI fragment and (after addition of EcoRI linker molecules to both ends) inserted into the EcoRI site of pGEM 1.
Biosynthesis of the transferrin receptors
pOEM 2dhfr
pOEM 2TR A 2 dhfr
.
pOEM 2g lobin
pOEM 2TR globin
A
4630 -
4m
am =.
__
18 4"W-
10-
do%w ..
S
S
S S MM M M PP
1 2 3 4 5 6 7
S
W. affl
S
S S MM M M M P P P .TX 8 9 10 11 12 1314 15 16 17 18 1920
Fig. 7. SRP dependence of hybrid proteins. The figure shows a fluorogram of SDS -PAGE analyses of [35S]methionine-labelled proteins obtained by translating fusion proteins in a wheat germ translation system in the presence or absence of salt-washed membranes (M) and SRP (S). Protease (P) and protease and Triton X-100 (TX) treated samples are indicated.
DOCKING PROTEIN
Fig. 8. Postulated biosynthesis of TR. Plasmid pGEM 2-dhfr (the dhfr sequence in the transcription plasmid). The DHFR coding sequence was obtained as a BamHI -HindIII fragment from the plasmid pDS5/3 (Stueber et al., 1984). This was directly inserted into the polylinker region of pGEM 2 to yield pGEM 2-dhfr (Figure lb). Plasmid pGEM 2-globin (the globin sequence in the transcription plasnid). Chimpanzee a-globin cDNA was excised as a NcoI - PstI fragment from a plasmid of Liebhaber and Begley (1983) as modified by Coleman (unpublished). The NcoI site was filled in and the fragment ligated into a HindU and PstI-restricted pGEM 2 plasmid (Figure lb). Plasmid pGEM 1 7RAI (codes for the TR molecule with its transmembrane peptide deleted). Deletion of the gene region encoding the transmembrane region
was done as follows. The pGEM 1 TR DNA was linearized at the unique BamHI site at the 5' side of the TR coding region (Figure ic). Using limited Bal31 exonuclease digestion, the TR coding sequence was removed from its 5' end toward the 3' end, to a point which was within or beyond the coding region for the TR transmembrane peptide ( - 340 bp). The digestion was monitored by following the size reduction of a suitable restriction endonuclease fragment. After endpolishing with T4 DNA polymerase and adding a ClaI linker molecule, the DNA was cleaved with ClaI and PvuI and the large fragments isolated. In order to regenerate the very 5' end of the TR coding region, which encodes the NH2-terminal cytoplasmic domain of this protein, we performed a three-fragment ligation as shown in Figure 4. The BamHI-HaeIIl/ClaI fragment contains the genetic information for the cytoplasmic domain. It was derived from a small 248-bp HaeIlI-HindIII fragment of pGEM TR which was ligated to ClaI linker molecules and then cleaved with BamHI. Deleted plasmids were first screened by restriction endonuclease analyses and those of interest were further checked for a preserved reading frame by in vitro transcription/translation. Final characterization was done by nucleotide sequence analyses. One plasmid, pGEM 1 TRAI was shown to have a suitable deletion (codons 55-91) encompassing the gene region for the transmembrane peptide. Plasmids pGEM 2 TR-dhjfr and pGEM 2 TR-globin (codes for TR-DHFR and TR-globin fusion proteins, respectively). Both plasmids contain a 333-bp EcoRI (positioned in the pGEM-1 polylinker region) -DdeI (position 550 in the TR map; Schneider et al., 1984) fragment derived from the TR cDNA in front of the reporter genes. The TR fragment encodes for the cytoplasmic tail and the transmembrane region of the TR followed by eight more amino acids. It was isolated from a larger 908-bp HindH -HindIl TR cDNA fragment which was first cleaved with DdeI and then, after having filled in the protruding ends with the E. coli polymerase I (Klenow fragment), with EcoRI (in the pGEM-1 polylinker). The EcoRI-DdeI fragment was isolated and inserted into plasmids pGEM-2 dhfr and pGEM-2 globin to give plasmids pGEM-2 TR-dhfr and pGEM-2 TR-globin respectively. The acceptor moleucles had been prepared as follows: pGEM-2 dhfr was cleaved with BamHI, the cohesive ends digested with Mung bean exonuclease, and finally
1549
M.Zerial et al. cut with EcoRI (in the pGEM 2 polylinker). Plasmid pGEM-2 globin was cleaved with NcoI, the protrudingfilled in, and then cut with EcoRI. A schematic representation of these constructions is shown in Figure Id. Plasmid pGEM 2TRA 2-dhfr (codes for aTRA 2-DHFR fusion protein; in this theTR? portion is restricted to essentially only its transmembrane peptide). This plasmid was derived from pGEM-2 TR-dhfr by performing a three-fragment recirculation ligation as shown in Figure le. Plasmid pGEM 2 TR-dhfr was first digested with Sacd and SphI yielding two fragments which were isolated. The shorter (1027 bp) fragment was cleaved with AluI. This produced two fragments of 492 and 535 bp. The latter fragment (containing two Haell site) was partially digested with HaeIH and the 387-bp HaeHI-SstI fragment was isolated. In vitro transcription and translation. Analyses of proteins In vitro synthesis of capped mRNA from the pGEM plasmids bySP6 polymerase was performed as described (Melton et al., 1984; Krieg and Melton, 1984; Konarska et al., 1984). Plasmids pGEM-1 TR and pGEM 1-TRA were the only linearized templates used. These had been cut with of plasmid DNA was PvuI before transcription. For RNA synthesis 0.3 5 incubated at 40°C with 3 units ofSP6 polymerase in a final volume ofyd and in the presence of 0.5 mM each of ATP, UTP and CTP, 0.1 mM GTP, 0.25 mM 7 mG(5')ppp(5')G, 5 units of RNase inhibitor, 2 mM spermidine, 100 /tg/ml BSA (DEPC-treated), 40 mM Hepes-KOH (pH 7.4) 6 mM magnesium acetate and 10 mM dithiothreitol (DTT). After 20 min incubation GTP was added to a final concentration of 0.2 mM and incubation continued for a further 20 min. In the control experiments DNA was substituted with TE. All buffers were prepared with DEPC-treated bidistilled water. Translation in the presence of reticulocyte lysate or wheat germ lysate in the presence or absence of RER membranes were done as described (Melancon and Garoff, 1986). TR and TRA-1 translation products were solubilized directly into 20 il of sample buffer for SDS gel analyses (80 mM Tris-Cl, pH 8.8, 0.40 M sucrose, 0.02% bromophenol blue, 4mM EDTA pH 8.0, 1% methionine, 3% SDS and 40 mM DTT) while those of DHFR, a-globin and the hybrid proteins were first precipitated in 10% trichloroacetic acid (TCA) at4°C for 30 min. All protein samples were heated to 95°C for 5 min and then alkylated with iodoacetamide. Proteins were analysed by either 8% or 15% PAGE, and procesed for fluorography as previously described (D.F.Cutler and H.Garoff, 1986). Protease treatments Globin and TR-globin translations were incubated with 0.5 mg/ml proteinase K at0°C for 1 h. TR and TRAI products were treated with 0.15 mg/ml protease also at 0°C for 1 h. DHFR or DHFR hybrids needed harsher conditions. These were incubated with 0.5 mg/ml protease at 300C for 30 min. Proteolysis was stopped by the addition of PMSF to a final concentration of 2 mg/ml. After a 5 min incubation on ice the samples were procesed for SDS gel electrophoresis. Immunoprecipitation ofTR Three 10-cm dishes with subconfluent HeLa cells were labelled with 250 ILCi of [35S]methionine (1000 Ci/mmol) for 2 h. Two dishes had been respectively for pre-incubated with tunicamycin at 4 Ag/ml and 12 2 h. After a 30 min chase in excess cold methionine the cells were lysed in 1mi of hot 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.4% SDS, 2 mM EDTA and 10,ug/ml PMSF. For immunoprecipitation 100 IL aliquots were used. These were made with Nonidet P40 and TR immunoprecipitated using 1 1l of a polyclonal antihuman TR antisera (Schneider et al., 1983) essentially as described in Cutler et al. (1986).
,ig
Ag/mli,
Acknowledgements We are grateful to Bernhard Dobberstein and David Meyer for advice and the generous gift of SRP and salt-washed rough microsomes. We wish to thank Hildegard Kern for expert technical help, Danny Huylebroeck for suggestions regarding DNA cloning, Annie Steiner for typing the manuscript and Perla Zerial for the drawings. Thanks are also due to Kai Simons for useful discussions and several members of the Cell Biology Program for critical reading of the manuscript. M.Z. was a recipient of EMBO and Fondazione Anna Villa Rusconi (Varese, Italy) short-term fellowships and P.M. of a short-term EMBL fellowship.
References Adams,G.A. and Rose,J.K. (1985a) Cell, 41, 1007-1015. Adams,G.A. and Rose,J.K. (1985b) Mol. Cell Biol., 5, 1442- 1448. Blobel,G. (1980) Proc. Natl. Acad. Sci. USA, 77, 1496-1500. Blobel,G. and Dobberstein,B. (1975) J. Cell Biol., 67, 852-862. Bos,T.J., Davis,A.R. and Nayak,D.P. (1984) Proc. Natl. Acad. Sci. USA, 81, 2327-2331.
1550
Cutler,D.F. and Garoff,H. (1986) J. Cell Biol., 102, 889-902. Cutler,D.F., Melancon,P. and Garoff,H. (1986) J. Cell Biol., 102, 902 -910.
Davis,N.G. and Model,P. (1985) Cell, 41, 607-614. Davis,N.G., Boeke,J.D. and Model,P. (1985) J. Mol. Biol., 181, 111- 121.
Dillon,J.-A.R., Nasim,A. and Nestmann,E.R. (1985) Recombinant DNA Methodology. John Wiley and Sons, NY. Friedlander,M. and Blobel,G. (1985) Nature, 318, 338-343. Garoff,H. (1985) Annu. Rev. Cell Biol., 1, 403-445. Grossman,A.R., Bartlett,S.G., Schmidt,G.W. and Chua,N.-H. (1980) Ann. N.Y Acad. Sci., 343, 266-274. Hargrave,P.A., McDowell,J.H., Curtis,D.R., Wang,J.K., Juszczak,E., Fong,S.-L., Mohana-Rou,J.K. and Argos,P. (1983) Biophys. Struct. Mech., 9, 235-248.
Hortsch,M. and Meyer,D.I. (1984) Biol. Cell, 52, 1-8. Konarska,M.M., Padgett,R.A. and Sharp,P.A. (1984) Cell, 38, 731-736. Krieg,P.A. and Melton,D.A. (1984) Nucleic Acids Res., 12, 7057-7070. Liebhaber,S.A. and Begley,K.A. (1983) Nucleic Acids Res., 11, 8915 -8929.
Maniatis,T., Fritsch,E.F. and Sambrook,J. (1982) Molecular Cloning, A Laboratory Manual. Cold Spring Harbor Laboratory Press, NY. Markoff,L., Lin,B.-C., Sveda,M.M. and Lai,C.-J. (1984) Mol. Cell Biol., 4, 8-16. Melancon,P. and Garoff,H. (1986) EMBO J., 5, 1551-1560. Melton,D.A., Krieg,P.A., Rebagliatti,M.R., Maniatis,T., Zinn,K. and Green,M.R. (1984) Nucleic Acids Res., 12, 7035-7056. Ovchinnikov,Y.A., Abdulaev,N.G., Feigina,M.Y., Artamonov,I.D., Zolotarev,A.S., Kostina,M.B., Bagachuk,A.S., Miroshnikov,A.I., Martynov,V.I. and Kudelin,A.B. (1982) Bioorg. Khim., 8, 1011-1014. Sabatini,D.D., Kreibich,G., Morimoto,T. and Adsnik,M. (1982) J. Cell Biol., 92, 1-22. Schneider,C., Asser,U., Sutherland,D.R. and Greaves,M.F. (1983a) FEBS Lett., 158, 259-264. Schneider,C., Kurkinen,M. and Greaves,M. (1983b) EMBO J., 2, 2259 -2263. Schneider,C., Owen,M.J., Banville,D. and Williams,J.G. (1984) Nature, 311, 675-678. Spiess,M. and Lodish,H.F. (1986) Cell, 44, 177-185. Stueber,D., Ibrahimi,I., Cutler,D., Dobberstein,B. and Bujard,H. (1984) EMBOJ., 3, 3143-3148. von Heijne,G. (1985) In Knauf,P.A. and Cook,J.S. (eds), Membrane Protein
Biosynthesis and Turnover. Academic Press,
NY, Vol. 24,
151- 179.
Walter,P., Gilmore,R. and Blobel,G. (1984) Cell, 38, 5-8. Watson,M.E.E. (1984) Nucleic Acids Res., 12, 5144-5164. Wickner,W.T. and Lodish,H.F. (1985) Science, 230, 400-406. Received on 22
April 1986
