Retroviral libraries with high-titer efficiency and larger insert potential
Katherine Felts Kim Zaharee Latha Sundar Jamie
Anna Waesche Peter Vaillancourt
We introduce a set of human cDNA expression libraries inserted into the high-titer retroviral vector pFB. ViraPort retroviral libraries allow complex cDNA libraries to be efficiently introduced into virtually any mitotic cell type for screening based on gene function. The cDNA copy number per cell can be easily controlled by adjusting the multiplicity of infection (MOI). In this way, cell populations may be generated in which greater than 90% of infected cells contain one-to-five cDNAs. We describe the isolation of two known oncogenes and one cell-surface receptor from one of the libraries (Burkitts Lymphoma).
Proliferation or survival
Loss of contact-inhibition
Loss of contact-inhibition
Loss of contact-inhibition
Survival: resistance to apoptosis inducers
In vivo metastasis
In vitro invasion
Phenotype: ion-specific indicator or tracer
Expression cloning allows genes to be identified and isolated based on gene function or phenotype in the absence of any prior knowledge of protein or nucleic acid sequence. A wide range of gene types have been cloned from complex cDNA libraries using functional assays that allow a specific trait to be selected or screened (Table 1).1 Originally, hybrid selection, direct transfection of plasmid expression libraries, and transfection of genomic DNA were successfully used to clone growth factors, cell surface proteins, and transcription factors. However, selected cells or cDNA pools usually contained a large number (often hundreds) of extraneous clones in addition to the target cDNA of interest, necessitating several rounds of laborious DNA isolation and rescreening before the gene of interest is identified.
The use of retroviral vectors for expression cloning has several advantages over traditional methods. Recent advances in viral packaging systems ensure that virtually any cell type can be transduced with efficiencies approaching 100%.2 Additionally, the copy number of individual cDNA expression cassettes can be easily controlled by varying the MOI. Hence, populations of infected cells may be generated in which greater than 90% of the cells are transduced with one-to-five individual cDNAs per cell, greatly reducing the time and labor required to isolate the gene of interest.3 The Moloney Murine Leukemia Virus (MMLV)-based vectors have a large insert capacity (8.0 kb).4 The retroviral vectors integrate into transcriptionally active regions of the host genome, and two days following infection, a collection of stable cells are produced in which there is minimal clonal variegation of expression from cell to cell. Finally, improved transfection methods have allowed transient production of high-titer viral supernatants containing representative mixtures of cDNAs without the concern of clonal skewing, which likely occurs with the production of stable producer cell lines.5
We have produced ViraPort retroviral cDNA expression libraries in the high-titer vector pFB6 (Figure 1A). This vector is an MMLV-based vector in which titers of 107 to 108 colony forming units (cfu)/ml are routinely achieved using 293 cell-based transient packaging systems. We validated the use of ViraPort libraries for functional cloning by screening the human Burkitts Lymphoma cDNA library for oncogenic cDNAs.
Polyadenylated mRNA was isolated from various human tissues or cell lines by two cycles of purification with oligo dT-cellulose. Double-stranded cDNA was synthesized using Stratagenes cDNA synthesis kit, and size-selected cDNA was inserted between the EcoR I and Xho I sites of the plasmid pFB (Figure 1A). The pFB vector has an extended packaging site and uses the viral splice donor and splice acceptor configuration from the high-titer, high-expressing vector pMFG.7 To minimize packaging size constraints, and because the high infection efficiency of the system obviates the need for selection of transduced cells, there are no selectable markers in this vector. To assess the quality of both the phosphatase-treated EcoR I/Xho I-digested vector as well as the quality of the Daudi cDNA, XL10-Gold ultracompetent cells were transformed with ligated plasmid. Miniprep DNA from 100 of the resultant colonies was digested and analyzed. All of the 100 clones examined had EcoR I/Xho I-excisable inserts, with an average size of 1.7 kb and a size range of 0.5 to 5.0 kb (data not shown). Primary clones (1.1 x 106) were amplified to construct the pFB Daudi library. Figure 2 shows the size distribution for 20 representative cDNAs from this library.
In the schematic showing a general strategy for functional cloning using ViraPort libraries (Figure 1B), high-titer viral stocks were transiently produced by transfecting a 293 cell-based retroviral producer line 8,9 with the plasmid cDNA library, then harvesting cell supernatants after 2 days. Alternatively, 293 or 293T cells (ATCC) may be cotransfected with the plasmid library in addition to one or more expression vectors that provide the viral structural proteins gag, pol, and env in trans.10,11 We used the former method for this study and achieved a titer of approximately 107 cfu/ml, as determined by RT-PCR using the prostar HF RT-PCR system12 and vector-specific primers. After an additional one or more days, cells were selected or screened for the desired phenotype, and colonies or selected populations were expanded and either enriched by one or more cycles of selection or analyzed directly. Genomic DNA was then isolated, and the cDNA inserts were retrieved by PCR using vector-specific primers that flank the insert. The PCR products were then subcloned into high-efficiency PCR cloning vectors (e.g., PCR-Script or pCMV-Script vectors) that allow sequencing of the insert and further functional analysis.
The ability of oncogene expression to confer deregulated growth on certain contact-inhibited cell lines, such as NIH3T3 fibroblasts, is a proven, powerful tool for identifying and isolating natural oncogenes, as well as genes involved in signal transduction pathways and cell cycle regulation.13 Constitutive expression or truncation of many of these latter classes of genes leads to a loss of contact inhibition and, thus, focus formation in NIH3T3 cells.
Insert size (kb)
Mitochondrial phosphate carrier
Gamma interferon-inducible protein
repressor of known cell-cycle regulatory protein
(to be discussed in a future article)
Human ribosomal protein L8
Unknown (pac clone; chromosome 11)
HL5 proteosome-associated protein
translation initiation factor
The pFB Burkitts Lymphoma (Daudi) library was screened for the presence of oncogenes by first infecting low-passage NIH3T3 cells, then isolating and expanding foci after 28 days. Uninfected NIH3T3 cells were maintained as a negative control. Focus formation was observed in both uninfected and infected cell populations. Approximately three times as many foci were observed on the infected plates, and, of those, a number of foci had unusual morphologies not observed on the control plates. Twenty-four of the largest and most aggressively growing foci on the test plates were picked and expanded. Genomic DNA was extracted from nine of the clones, and cDNA inserts were PCR-amplified, subcloned, and sequenced. Due to the relatively high MOI used in this screen, none of the foci had less than four individual cDNA inserts, and one of the foci (Onco-11, Table 2) had as many as nine individual cDNAs. Seven of the nine foci examined harbored cDNAs encoding N-terminally truncated versions of the signaling proteins raf-1 and vav-1, which have been shown to be potent transforming proteins when truncated at their N-termini.13. Onco-1 contained a gene encoding a transcriptional repressor of a known cell cycle regulatory protein. It is not surprising, therefore, that constitutive expression of this cDNA might lead to unregulated cell growth. Onco-11 contained nine distinct cDNAs, none of which were obvious candidates for oncogenic transformation, and three of which were unknown. Like Onco-1 and Onco-11, all seven of the raf-1 and vav-1 foci harbor at least three additional cDNAs (not shown in Table 2). These additional cDNAs were distinct from each other and from focus to focus, an observation expected for a complex, representative cDNA library for which no selection applies other than that for the single focus-forming oncogene.
It is possible to clone cell-surface receptors by expression screening, provided that a receptor ligand or receptor-specific antibody is available. We screened for the IL-4 receptor (IL-4 R), which is known to be expressed on the surface of Daudi cells. NIH3T3 cells were infected at an MOI of approximately one, and, after 2 days, the cells were labeled with receptor-specific monoclonal antibody followed by a fluorescent anti-mouse IgG secondary antibody. The cells were then FACS sorted, and the sorted population was expanded for 10 days and sorted a second time. Approximately 3% of the expanded population from the first sort was collected in the second sort (Figure 4A) and expanded an additional 10 days. FACS analysis of the twice-sorted population indicated that over 90% of the cells reacted with the receptor-specific antibody. PCR reactions were performed using vector-specific primers, which flank the cloning site, and genomic DNA was isolated from either uninfected NIH3T3 cells or from the sorted population as template. Figure 4B shows a prominent 1.7-kb band that was observed for the sorted population (Lane 3), whereas no PCR product exists for the uninfected control (Lane 2).
The 1.7-kb PCR product was subcloned, sequenced, and found to encode not the IL-4 R but the human IgG FcaRII receptor, a receptor found on most hematopoietic cell types that recognizes the Fc domain of IgG antibodies.14 Although the a-IL-4 R monoclonal antibody used in the FACS sort was derived from mouse cells, it is not surprising that the human IgG Fc receptor can recognize the Fc region of murine IgG because there is a high degree of homology between both the murine and human IgG constant regions, as well as between their respective IgG FcaRII receptor chains.14 Thus, the a-IL-4 R antibody functioned in this screen not as an IL-4 R-specific antibody but as an IgG FcaRII-specific ligand. This serves to better illustrate the power of FACS for receptor cloning because, when screening for an unknown receptor, it is more likely to have at hand receptor-specific ligand rather than receptor-specific antibody.
ViraPort retroviral libraries can be used in a wide range of applications and are limited only by the ability to devise a functional screen for the desired target gene. In addition, libraries comprising a complex spectrum of mutants derived from a single gene can be screened for dominant positive or negative mutations or for other altered or enhanced phenotypes.1 The availability of packaging systems that use polytropic envelope proteins, such as 10A1 and VSV-G, will allow functional screening in essentially any mitotic cell type.2 Finally, the relatively high titers and high expression levels that can be attained using the pFB vector backbone6 increase the probability of being able to identify the desired cDNA from a complex, fully representative library.
The authors thank Dr. Gary P. Nolan (Stanford), Dr. John Wu (Berlex), Mei Vaillancourt (Canji), Dr. Mary Simcox (Hoffman-La Roche), Dr. Ronda Allen, Keith Chen, Tanya Hosfield, Milan Miletic, John C. Bauer, Dr. Alan Greener, Dr. Joe Sorge, Dr. Cathy Chang, Tim Sanchez, Connie Hansen, Elliott Gorfain, and Mike Kobrin for helpful discussions and technical assistance.
Kitamura, T. (1998) Intern. J. Hematol. 67: 351-359.
Miller, A.D. (1997). In Retroviruses (Eds. Coffin, J.M., Hughes, S.H., and Varmus, H.E.) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. pp. 437-473.
Onishi, M., et. al. (1996) Exp. Hematol. 24: 324-329.
Riviere, I and Sadelain, M. (1997) In Methods in Molecular Medicine: Gene Therapy Protocols (Ed. Robbins, P.D.) Humana Press, Totawa, New Jersey. pp. 59-78.
Kitamura, T., et. al. (1995) Proc. Natl. Acad. Sci. USA 92: 9146-9150.
Felts, K., et. al. (1999) Strategies 12: 74-77
Riviere, I., et. al. (1995) Proc. Natl. Acad. Sci. USA 92: 6733-6737.
Pear, W.S., et. al. (1997) In Methods in Molecular Medicine: Gene Therapy Protocols (Ed. Robbins, P.D.) Humana Press, Totawa, New Jersey. pp. 41-57.
Kinsella, T.M. and Nolan, G.P. (1996) Hum. Gen. Ther. 7: 1405-1413.
Soneoka, Y., et. al. (1995) Nuc. Acids Res. 23(4): 628-633.
Yang, S., et. al. (1999) Hum. Gen. Ther. 10: 123-132.
Borns, M., et. al. (1999) Strategies 12: 33-36.
Hesketh, R. (1997) The Oncogene and Tumour Suppressor Gene Facts Book. Academic Press, San Diego.
Brooks, D.G., et. al. (1989). J. Exp. Med. 170: 1369-1385.