December 1997
The Construction of a Bait for Use in a Yeast Two-Hybrid System to Screen for Protein-Protein
Interactions with GlcNAc-TV
By ERIC STRONG
Massachusetts Institute of Technology
|
|
|
Abstract
GlcNAc-TV is an important glycosyltransferase that acts as a modifier of membrane bound glycoproteins by catalyzing the transfer of N-acetylglucosamine from
UDP-N-acetylglucosamine to a-6-D-mannoside. It has been previously shown that an increase in the b1,6 branch sites of complex asparagine linked
oligosaccharides on glycoproteins that is characteristic of GlcNAc-TV activity is correlated with an increase in metastatic potential in human prostate tumors. In this
paper, we report that a bait containing sequence coding for the catalytic domain of GlcNAc-TV was constructed for use in a yeast two-hybrid system. This bait was
designed to screen cDNA libraries derived from both normal and metastatic prostate tissue in order to identify proteins which may interact directly with GlcNAc-TV.
It is hoped that by identifying such proteins, a better understanding of metastasis within the prostate may be achieved, ultimately leading to the discovery of new
markers for increased metastatic potential.
Introduction
Mammalian cells rely heavily on a diverse repertoire of complex glycan units attached to proteins for a variety of functions, including protein targeting and cell-cell
signaling. a-1,3(6)-mannosylglycoprotein-b-1,6-N-acetylglucosaminyltransferase (GlcNAc-TV) is one of a series of enzymes that act as post-translational modifiers
on membrane bound proteins by transferring and altering oligosaccharides bound to these proteins. GlcNAc-TV specifically transfers N-acetylglucosamine from the
nucleotide sugar, UDP-N-acetylglucosamine, to a-6-D-mannoside, forming b1,6 branch sites on asparagine (N) linked oligosaccharides on proteins containing the
sequence : Asn-X-Ser/Thr, where X is any amino acid other than proline1,2. Warren3 and Smets4 have
previously shown that malignant transformation within mammalian cells is associated with the gain of N-linked oligosaccharides. Specifically, an increase in the
number of b1,6 branch sites on individual proteins has been correlated with an increase in metastatic potential, and studies suggest that oligosaccharides containing
the b1,6 branch site may be an important factor in the conditions necessary for the efficient metastatic spread of cancer cells5,6,7. This relationship has been demonstrated by experiments showing that a reduction in the formation of N-linked structures by somatic
mutations8,9,10 and drugs11 leads to a reduction in the metastatic potential of human melanoma
cells and leads to slower tumor growth in mice. Also, a specific upregulation of GlcNAc-TV by transfection can lead to an increase in cell motility and aggressiveness
without requiring additional mutations12.
The yeast two-hybrid system is an excellent molecular approach to study interactions between GlcNAc-TV and other membrane proteins which may play a role in
metastasis. The two-hybrid system is composed of three main components13,14. The first component is the "bait", a
fusion of a known protein of interest and a DNA binding domain. The second component is a "reporter" gene whose expression is easily detected and which lies
downstream from the promoter region recognized by the bait. The third component is known as the "prey", which is a second fusion, one between an unknown
protein and an activation domain compatible with the specific reporter system. Thus, expression of the reporter gene only occurs when the prey and bait interact and
bring an activation domain to the immediate vicinity of the promoter. By creating a cDNA library of possible preys, the library can be screened for positive bait/prey
interactions by selecting for expression of the reporter gene. Yeast colonies positive for this expression can then be isolated, and the gene within the interacting prey
can be studied more thoroughly.
To study the increased metastatic potential of human prostate tumors, we will screen cDNA libraries derived from both normal prostate tissue and a metastatic
prostatic cell line with baits containing sequence encoding for a portion of GlcNAc-TV. Once interacting proteins are discovered, their encoding genes can be
isolated, identified, and further characterized to increase understanding of metastasis within the prostate. Also, by finding other enzymes in the asparagine linked
oligosaccharide processing pathway that interact directly with GlcNAc-TV or by identifying membrane proteins which are modified directly by GlcNAc-TV, new
markers for an increase in metastatic potential may be found.
Materials and Methods
Analysis of Clone Set A
Various clones of Escherichia coli transformed with cloning vector pEG202 containing inserts derived from human GlcNAc-TV cDNA were provided by Sandra
Gaston (MIT). pEG202 encodes a lexA protein (DNA binding domain) which lies upstream of a multicloning site and is under the control of an ADH1 promoter.
pEG202 also encodes for the his3 gene so that it can be selected for in his- media. These previously constructed clones were created from the following ligation
reactions: GlcNAc-TV primed with 2F (GCATGAAGAATTCCGGTGGATGAGA), (nt 838-863) and 7R (CCTCTACTTCCTCCTGATTGTTGAG),(nt
1910-1886)
(Fig.1),
and ligated into
pEG202 vector using the vector and inserts' EcoRI and PstI sites at a high relative concentration of insert ; two separate
reactions of GlcNAc-TV primed with 2F and 7R, and ligated into pEG202 using the EcoRI and PstI sites at a low relative insert concentration ; GlcNAc-TV primed
with 1F (TCCGTGGAAGTTGTCCTCT),(nt 160-178) and 7R and ligated into pEG202 using only the EcoRI site at a low relative insert concentration. Plasmids
were isolated from 20 colonies from each reaction using Qiagen prep kits (Qiagen, Inc.). The inserts of these plasmids were amplified using PCR. Amplification was
carried out using 30 cycles of denaturing for 30 sec at 92°, annealing for 2 min at 60°, and extension for 30 sec at 75°, with a final extension of 5 min. The following
primers flanking the insert were used: Bait F (CGTCAGCAGAGCTTCACCATTG) and Bait RB (TTCGCCCGGAATTAGCTTGGC) for the first three reactions,
Bait F and Bait R PstI (GGAATTAGCTTGGCTGCAG) for the fourth reaction. Clones displaying clean amplification products when run on a 1.5% agarose gel
were prepped using miniprep columns (Qiagen, Inc.). These minipreps were then used in a series of diagnostic restriction digests with HindIII, EcoRI, and SphI, and
were sent to the HHMI Biopolymers Laboratory at MIT to be sequenced.
Construction of Clone Set B
PCR reactions were completed using a cDNA library derived from normal prostate tissue (Clonetech) as the template. Amplification was carried out using 30 cycles
of denaturing for 30 sec at 92°, annealing for 2 min at 55°, and extension for 30 sec at 75°, with a final extension of 5 min. The primers used were 2F and 5R
(TTTCTCGAAGAAGGAACTGCAGGTCCG),(nt 1670-1644). This PCR product was digested with EcoRI and PstI, and ligated into pBluescript II KS vector
that had already undergone a sequential digested by PstI and EcoRI. The ligation reaction consisted of a 3:1 molar ratio of insert to vector. These new clones were
then transformed into chemically competent E. coli (provided by Tom McHugh, MIT). 2ml of the 20ml ligation reaction were added to 50ml of these cells, they were
incubated on ice for 20 min, heat shocked at 42° for 45 sec, incubated again on ice for 2-3 minutes, and incubated for 60 min in 900ml of LB before being plated.
pBluescript II KS is a cloning vector whose multicloning site is contained within a lacZ structural gene. Thus, transformants that contain positive ligation products can
be identified by their white color when plated on X-gal in the presence of IPTG. When positive transformants were identified, their inserts were reamplified directly
out of the bacteria by PCR. Colonies were picked and each added to 50ml of cracking buffer (0.4ml 0.5M EDTA, 1ml Triton X-100, 2ml 1M Tris pH 8.0, 96.6ml
water), before they were boiled and centrifuged at top speed for 5 min. Plasmid contained in the supernatant was then used as the template for the reactions whose
conditions were as follows: annealing temperature of 52° and multiclonal site flanking primers T7 (GTAATACGACTCACTATAGGGC) and T3
(AATTAACCCTCACTAAAGGG). These ligation products were cleaned by spin columns (Qiagen, Inc.) before they were sent to the HHMI Biopolymers
Laboratory at MIT to be sequenced.
Construction of Clone Set C
PCR reactions using the primer pairs 1F/6R, 2F/5R, 11F/11RG6, 10F/7R, and 4F/7R were run using the plasmid pCDNA3-FLHuTV (provided by Michael Pierce,
Univ. of Georgia) as a template (Fig.1). This plasmid contains the entire GlcNAc-TV sequence from human liver cloned into the multiclonal site of the expression
vector pCDNA3 (Invitrogen, Inc.). These PCR reactions were carried out using 30 cycles of denaturing for 30 sec at 92°, annealing for 2 min at 55°, and extension
for 1 min at 75°, with a final extension of 7 min. The products of the reactions using the 1F/6R and 2F/5R primer pairs were ligated into pEG202 using the multiclonal
restriction sites EcoRI and BamHI, and EcoRI and NcoI, respectively. These ligation products were transformed into chemically competent E. coli as described
above and selected for ampicillin resistance. The pEG202 plasmids from these transformants were used as a template in PCR reactions completed directly from
bacteria as described above for the purpose of identifying desired constructs. The reactions were carried out using the primers Bait F and Bait RB with 30 cycles of
denaturing for 30 sec at 92°, annealing for 2 min at 55°, and extension for 1 min at 75°, with a final extension of 7 min.
The ligation products from above were again transformed into chemically competent E. coli cells using the same transformation protocol with the modification of
using 4ml of the ligation reaction in the transformation reaction instead of the previous 2ml. PCR reactions using the Bait F and Bait RB primers were completed on
transformants as described just previously, and amplification products which were of the expected size were used in a diagnostic restriction digest using the enzyme
ApoI.
Transformation of Bait
The newly constructed bait plasmids containing correctly oriented GlcNAc-TV fragments were prepped using maxiprep kits (Qiagen, Inc.). The plasmids were then
transformed into the yeast strain EGY48 (ura3, his3, trp1, LexAop6-leu2), which had been previously transformed with the reporter plasmid pSH18-34. This
plasmid has a lacZ gene unding a LexA promoter and a ura3 gene so its presence can be selected for in ura- media. Successful transformation of pSH18-34 into
EGY48 was then selected for by plating on ura- plates, and successful transformation of the bait plasmids into EGY48+pSH18-34 was selected for by plating on ura-
his- plates. To transform the bait and reporter plasmids into EGY48, a colony of EGY48 was innoculated overnight in 5ml of the appropriate liquid dropout media.
The cells were spun down at 3000rpm for 4 min, resuspended in 1ml water, spun down briefly, resuspended in .5ml TE/LiOAC (5ml 1M LiOAc pH 7.2, 0.5ml 1M
Tris-HCl pH 7.5, 0.1ml .5M EDTA pH 8.0, 44.4ml water), and spun down briefly again. The cells were then resuspended in TE/LiOAc before adding the following
to each 50ml if cells: 5ml of boiled salmon sperm, 1mg plasmid DNA, 6ml DMSO, 300ml PEG/TE/LiOAc (4ml 50% PEG, 500ml 1M LiOAc pH 7.2, 50ml 1M
Tris-HCl pH 7.5, 10ml .5M EDTA pH 8.0, 440ml water). The transformation mixture was the incubated at 30° for 30 min, heat shocked at 42° for 15 min, washed
in 1ml of water, spun down, and resuspended in water to a final volume of 300ml prior to plating.
Results
Characterization of Clone Set A
After the characterization of clone set A, we found that none of the examined clones contained the expected sequences of the GlcNAc-TV gene. Of the 80 colonies
containing various preconstructed plasmids which were examined, we were able to amplify fragments from 30 colonies
(Fig.2).
Of these 30 fragments, only 2 were of
the expected size, and most were much smaller (~.2kb). From these 30, 15 clones were chosen to be characterized further. These 15 were purified using the Qiagen
procedure for plasmids, and were then analyzed using a restriction digest by HindIII. This digest produced unexpected results. Based on a restriction map of the
vector plasmid (pEG202), three bands were expected: 5.15kb, .18kb, and a variable band of length 4.47kb + the length of the insert. Only two clones (E3 and E4)
were consistent with the predicted restriction map derived from the published GlcNAc-TV sequence derived from a human cDNA library15.
To characterize in greater detail, 9 of the 15 clones were digested separately by EcoRI and SphI and visualized on a 1.5% agarose gel. These digests also produced
results that were inconsistent with the predicted restriction map, and few bands were of the anticipated size. The cloned inserts of these nine clones were then
sequenced. The sequencing results showed that all nine clones lacked any GlcNAc-TV fragments, and the plasmid region which had been assumed to contain these
fragments only contained various segments of vector sequence. These clones were then set aside, and no further analysis was carried out on them.
Construction of Clone Set B
After we amplified a GlcNAc-TV fragment out of the cDNA library from normal prostate tissue using the 2F/5R primers, and ligated it into the pBluescript II KS
vector, we transformed these new clones into E. coli. Successful insertions were selected by use of a blue/white screen. Nine white colonies were further
characterized, along with one blue colony as a control. After the insert was reamplified out of the Bluescript plasmid, it was run on a 1.5% agarose gel
(Fig.3). Four
of the white colonies displayed a strong band at the expected size of ~1kb. An additional white colony displayed a slightly larger band at ~1.5kb. It was
hypothesized that this larger band might be the result of a multiple transcript of the GlcNAc-TV gene.
To provide evidence for this hypothesis, these PCR products were then sequenced. However, sequencing data showed these products to only contain sequence
from pJG4-5, the plasmid used in construction of the cDNA library. It was then considered that such unexpected ligation products might have arisen from a possible
polymorphism within the GlcNAc-TV EcoRI site used in ligation, disrupting EcoRI's ability to recognize the desired restriction site. This consideration was based on
the observation that this particular site is not conserved between human and Chinese hamster16. A base pair that is not conserved between
species is less likely to be conserved within a species. However, sequencing of the original 2F/5R PCR product which was digested prior to the ligation reaction
showed that this product also contained only pJG4-5 sequence. The 2F/5R primers may have misprimed due to non-optimal annealing conditions, but further
investigation with this clone set was not continued.
Construction of Clone Set C
After sequencing the 2F/5R bait from above revealed no GlcNAc-TV, construction of a new bait was begun. PCR reactions with various primer pairs were repeated
using pCDNA3-FLHuTV as the template as described above. All five PCR reactions displayed the correct band sizes when run on a 1.5% agarose gel
(Fig.4). The
PCR products primed with the 2F/5R and 1F/6R were chosen to be cloned into pEG202. These clones were then transformed into E. coli. Six colonies were
examined: four derived from the 2F/5R product and two from the 1F/6R product. After PCR amplification of the pEG202 inserts from these transformants, the PCR
products were run on a 1.5% agarose gel
(Fig.5).
Only one transformant appeared to contain an insert,however, based on its size, the GlcNAc-TV sequence, and a
diagnostic restriction digest with AluI
(Fig.6),
it was consistent with an insert of a segment of GlcNAc-TV from bp 684 to bp 847 that had been cloned into pEG202
in the incorrect orientation for a functional bait. Sequencing the insert and its flanking regions confirmed this hypothesis.
Once the sequencing had demonstrated this transformant did not contain a useful bait, transformation into E. coli was reattempted uding a two-fold increase in the
amount of plasmid added to the transformation reaction. Once again, the pEG202 plasmid from the transformants was used as the template in a PCR reaction
designed to amplify back out the insert. This set of PCR reactions produced two products which were of a desired size, both 900-1000bp
(Fig.7). Based on the
GlcNAc-TV sequence, it was thought that these may have contained a correctly oriented and in-frame fragment of GlcNAc-TV from bp 847 to bp 1626. A
diagnostic restriction digest with ApoI showed that both inserts produced the ~400bp and ~500bp fragments expected from this stretch of GlcNAc-TV sequence,
supporting the conclusion that a potentially viable bait had been constructed
(Fig.8).
Sequencing by the HHMI Biopolymers Laboratory confirmed that this was the
fragment that had been inserted into pEG202.
Discussion
Both restriction digests and sequencing data suggest that within clone set C, we successfully constructed a yeast two-hybrid bait for identifying protein-protein
interactions with GlcNAc-TV. In the construction of our bait, the suspected function of the region of GlcNAc-TV which this bait encoded was a great concern.
Because we were primarily interested in identifying proteins which were modified by GlcNAc-TV, we designed our bait (bp 847 - bp 1626) such that it encoded for
what we suspect to be a portion of the catalytic domain of the protein, specifically, amino acid residues 235 to 493. This suspicion was based on the known physical
characteristics of the GlcNAc-transferase family. All cloned members of this family to date share a common topology: a short N-terminal cytoplasmic tail, a
hydrophobic signal-anchor transmembrane domain, an extended stem region, and a large C-terminal catalytic domain17,18.
Despite GlcNAc-TV's predicted molecular weight of 85kDa, Weinstein16 reported bands as low as 69kDa which still maintained measurable catalytic
activity when purifying preperations of GlcNAc-TV. These smaller forms of GlcNAc-TV are the likely result of proteolytic cleavage in the stem region, separating the
transmembrane and catalytic domains to form a soluble, secreted form of the enzyme, similiar to the behaviors noted in other glycosyltransferases19,20.
The finding that this enzyme can lose ~200 amino acid residues from its N-terminus and still retain catalytic function suggests that our
bait's amino acid sequence from residues 235 to 493 is completed contained within the globular catalytic domain. Despite the observation that most
glycosyltransferases are encoded by five to six exons, GlcNAc-TV is much larger and more subdivided, as it is encoded by sixteen exons21. Our
bait includes exons six through eleven. Our GlcNAc-TV bait also contains three possible glycosylation sites at residues 334, 433, and 447.
Due to the great uncertainty regarding specific structural components of glycosyltransferases, we find it difficult to predict what other proteins might be found to
interact with our bait. Glycosyltransferases may be similiar to lectins in that the glycoconjugate binding domains might consist of pockets made up of short conserved
regions separated by much larger, variable ones17. Our GlcNAc-TV bait might then contain either small UDP-GlcNAc binding sites, acceptor sugar
binding sites, or both. Whether these sites are near one another or completely separated is still unknown. If our bait contained an acceptor sugar binding site, a
cDNA library screen using this bait might isolate those glycoproteins which are specifically acted upon by GlcNAc-TV.
Other possible proteins which might interact with our GlcNAc-TV bait are other transferases which act immediately before or after GlcNAc-TV in the processing
pathway of N-glycans, or other transferases which compete with GlcNAc-TV for the b1,6 linked mannose substrates. These include GlcNAc-TII, GlcNAc-TIII,
b1,4-galactosyltransferase, and a2,6-sialyltransferase2. Although GlcNAc-TI acts just prior to GlcNAc-TV in the sysnthesis of complex N-glycans,
Weinstein16 has previously reported that the two enzymes do not appear to directly interact at a significant proportion.
We now plan on using our GlcNAc-TV bait to screen various cDNA libraries derived from both normal and metastatic prostatic cell lines. Once proteins which
interact directly with GlcNAc-TV are discovered, they can be isolated, identified, and further characterized to increase our understanding of metastasis within the
prostate.
Acknowledgements
The author wishes to thank Rita Leung, Sharon Shen and WenYee Tsai for their support and advice throughout this project.
References
| 1. | Lodish H, et al. Molecular Cell Biology,3rd ed. 1995. Scientific American Books, Inc. |
| 2. | Fukuda, M. "Cell surface carbohydrates: cell-type specific expression". Molecular Biology Oxford University Press. 1994. p1-52. |
| 3. | Warren L, Buck CA, Teuszynski GP. "Glycopeptide changes and malignant transformation: a possible role for carbohydrates in malignant behavior". 1978. Biochem Biophys Acta 516:97-127. |
| 4. | Smets LA. "Carbohydrates of the tumor cell surface". 1984. Biophys Acta 738:237-249. |
| 5. | Fernandes B, et al. "1-6 branched oligosaccharides as a marker of tumor progression in human breast and colon neoplasia". 1991. Cancer Res 51:718-723. |
| 6. | Dennis JW. "Asn-linked oligosaccharide processing and malignant potential". 1988. Cancer Surv 7:573-595. |
| 7. | Dennis JW, et al. "1-6 branching of asn-linked oligosaccharies is directly associated with metastasis". 1986. Science 236:582-585. |
| 8. | Dennis JW. "Different metastatic phenotypes in two genetic classes of WGA-resistant tumor cell mutants". 1988. Cancer Res 46:4594-4600. |
| 9. | Finne J, Burger MM, Prieels JP. "Enzymatic basis for a lectin resistant phenotype: increase in a fucosyltransferase in mouse melanoma cells". 1982. J Cell Bio 92:277-282. |
| 10. | Ishikawa M, Dennis JW, Kerbel, RS. "Isolation and characterization of spontaneous wheat germ agglutinin-resistant human melanoma mutant displaying remarkable different metastatic profiles in nude mice". 1988. Cancer Res 48:665-670. |
| 11. | Dennis JW. "Effects of swainsonineand polyinosinicpolycytidylic acid on murine tumor cell growth and metastasis". 1986. Cancer Res 46:5131-5136. |
| 12. | Demetriou M, et al. "Reduced contact-inhibition adn substratum adhesion in epithelial cells expressing GlcNAc-transferase V". 1995. J Biol Chem 130(2):383-392. |
| 13. | Phizicky EM, Fields S. "Protein-protein interactions: methods for detection and analysis". 1995. Microbiological Reviews 59:94-123. |
| 14. | Finley RL, Brent R. "Interaction trap cloning with yeast". 1996. DNA Cloning and Expression Systems: A Pracitcal Approach. Oxford University Press. 1996. p169-203. |
| 15. | Nishikawa A, et al. "cDNA cloning and chromosomal mapping of human N-acetylglucosaminyltransferase V+". 1994. Biochem Biophys Res Commun 198(1):318-327. |
| 16. | Weinstein J, et al. "A point mutation causes mistargeting of golgi GlcNAc-TV in the lec4A Chinese hamster ovary glycosylation mutant". 1996. J Biol Chem 271:27462-27469. |
| 17. | Joziasse DH. "Mammalian glycosyltransferases: genomic organization and protein structure". 1992. Glycobiology 2:271-277. |
| 18. | Paulson JC, Colley KJ. "Glycosyltransferases: structure, localization, and control of cell type-specfic glycosylation". 1989. J Biol Chem 264:17615-17618. |
| 19. | Weinstein J, et al. "Primary structure of galactoside-2,6-sialytransferase". 1987. J Biol Chem 262:17735-17743. |
| 20. | Colley KJ, et al. "Conversion of a Golgi apparatus sialyltransferase to a secretory protein by replacement of the NH2-terminal signal anchor with a single peptide". 1989. J Biol Chem 264:17619-17622. |
| 21. | Saito H, et al. "Organization of the human N-acetylglycosaminyltransferase V gene". 1995. Eur J Biochem 233:18-26. |
|
|
|
