| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
Inst. für Arterioskleroseforsch. an der Univ. Münster, Domagkstr. 3, 48149 Münster, and
2
Inst. für Klin. Chem. und Laboratoriumsmed., Univ. Münster, Albert-Schweitzer-Str. 33, 48149 Münster, Germany;
a author for correspondence: fax 49-251-83 6208, e-mail cullen{at}uni-muenster.de
Although not yet a part of clinical chemical practice, analyzing patterns of differential gene expression in tissues such as cancers will likely become a routine method in coming years. The technique of PCR-based differential display developed by Liang and Pardee (1) has become increasingly popular during the last 4 years and is a considerable advance on conventional methods used to identify differentially expressed genes, such as differential hybridization and subtractive library construction. The advantages of differential display are the low RNA requirement and high versatility. However, serious problems remain, one of the most important being the high rate of false positives. Several methods have been described to deal with this problem, including the simultaneous display of PCR products from two uninduced and two induced lines with the requirement that the patterns from the pairs of uninduced or induced lines agree (2); the running of duplicate, identical samples from each RNA preparation side by side (3)(4); the display of PCR products from uninduced and induced lines over a time course of induction (2); and the repeating of experiments for those lanes in which putative candidate bands were identified (5).
We describe here a method for effectively eliminating false positives and spurious true positives (i.e., genes that really are regulated but whose regulation is not a result of the manipulation under investigation). This method is particularly useful when investigating gene expression in nonstandardized material such as fresh human or animal tissue or cells.
We use PCR-based differential display to investigate gene regulation in
human foam cells produced by loading macrophages for 48 h with 80
µg/mL acetlyated (Ac) or oxidized (Ox) low-density lipoprotein (LDL).
Human peripheral blood monocytes are isolated from volunteers by a
combination of cell separation and countercurrent elutriation as
previously described (6). Aliquots of the fractions are
examined for their purity in a FACScan (Becton Dickinson, San Jose,
CA). Fractions containing >95% monocytes are pooled and the cells are
plated at a density of 109 cells/L in 35-mm cell culture
dishes (Falcon, Heidelberg, Germany) in RPMI 1640 (Gibco, Eggenstein,
Germany) and incubated at 37 °C in a humidified incubator (5%
CO2). After 1 h, the monocytes adhere to the surface
of the dishes, and nonadherent cells are removed by washing. The
monocytes are then cultured for 14 days in RPMI 1640 supplemented with
200 mL/L human serum obtained from healthy volunteers. Total cellular
RNA from unloaded (control) macrophages and from macrophages that have
been loaded with cholesterol by using either AcLDL or OxLDL is isolated
by means of a standard procedure using guanidinium isothiocyanate
(7), and differential display is performed with a
commercially available kit (Display Systems, Los Angeles, CA) and a
standard protocol using AmpliTaq DNA polymerase (Perkin-Elmer, Norwalk,
CT) and 
-[35S]dATP (Amersham, Middlesex, UK).
To identify spurious changes in the band pattern (false positives), we perform all experiments (including controls) twice, i.e., we prepare RNA from two different batches of macrophages isolated separately from different volunteers and loaded with different preparations of modified lipoproteins. We then load the corresponding amplified cDNA products from both differential display PCR reactions in adjacent lanes on a 6% denaturing polyacrylamide sequencing gel 53 cm long. After 3 h, the gel is carefully transferred from the glass plates of the gel chamber to blotting paper and thoroughly dried at 80 °C for 1 h by using a vacuum gel-drier (Bio-Rad, Munich, Germany). This last step ensures that the dried gel is completely flat, because it is difficult, if not impossible, to achieve perfect alignment of a gel that has buckled.
After electrophoresis, the gels are exposed to Kodak XAR-5 film for
48 h. Achieving exact alignment of the dried differential display
gel with the exposed x-ray film is difficult, but we eliminate this
problem by using TrackerTapeTM (Amersham), a
phosphorescent tape that can be written on and then used to label the
dried gel and hence the corresponding x-ray film. We mark the
TrackerTape with suitable symbols and stick it onto opposite corners of
the gel (Fig. 1
) before exposure. By using the symbols written on the tape, we
can exactly align the developed film with the gel. To mark the
differentially exposed bands on the dried sequencing gel, we punch
through the four corners of each band of interest in the film with a
pointed scalpel.
|
A representative differential display pattern from a typical double
experiment is shown in Fig. 2
. The amplification products of the RNA preparations from two
different batches of control cells are shown in lanes 1 (donor 1) and 2
(donor 2), and the products of the PCR using the same primers on RNA
preparations from the same two different batches of cells in the loaded
state are shown in lanes 3 (donor 1) and 4 (donor 2) from AcLDL-loaded
cells and lanes 5 (donor 1) and 6 (donor 2) from OxLDL-loaded cells.
Typically, 120–150 bands are seen per lane.
|
Most bands are present in all six lanes, which provides an important check on the reaction in general: Gross differences between the lanes are likely to represent a fundamental problem with the reaction (e.g., sample mix-up). Arrow A indicates a band that, although almost absent in all four of the loaded cell preparations (lanes 3–6), is present in only one of the unloaded cell preparations (lane 1). This band is therefore a false positive (for downregulation of the mRNA in question) but would have been classified as a true positive if the experiment had not been performed in duplicate, i.e., if only cells from donor 2 had been used as control.
Arrow B indicates a band that is present in both the unloaded (lane 1) and loaded (lanes 3 and 5) cells from donor 1 but absent in both the unloaded (lane 2) and loaded (lanes 4 and 6) cells from donor 2 (spurious true positive). In our system, we have found that about one quarter of all differentially expressed genes are of this type, i.e., they are differentially expressed from cell preparation to preparation, but not in response to the manipulation under investigation (in our case, loading of the cells with AcLDL or OxLDL). The importance of this phenomenon is that these genes are not false positives because they are truly differentially expressed. Thus, if our experiment had been performed with unloaded cells from donor 1 and loaded cells from donor 2 (or vice versa), we would have come to the incorrect conclusion that expression of the band shown is related to cholesterol-loading of the cells. Furthermore, because this is a truly differentially regulated gene, we would have confirmed this incorrect result by using Northern analysis.
Arrow C indicates a band that is present in both of the control preparations (lanes 1 and 2) and absent in both of the preparations from the loaded cells (lanes 3–6, true positive). We cut out such bands for further analysis. In our experience, it is usually not worthwhile isolating bands from the lower 1/3 to 1/2 (lower ~20 cm) of the gel because these are often short 3' cDNAs that are difficult to characterize because 3' ends of known genes are often absent from the genetic databases.
Performing all experiments in duplicate using different cell donors and different sets of reagents has three advantages: (a) The incidence of false positives is greatly reduced by allowing identification of genes that are consistently regulated by the manipulation under study. (b) It avoids the isolation of genes that are truly differentially regulated but whose regulation is not related to the manipulation under study but to other factors specific to the individuals under study. (This is particularly important when experiments are performed with nonstandardized systems such as freshly isolated human or animal cells or tissues. We have not pursued genes that are differentially regulated between various cell types and we are not aware of the reasons for the differential regulation. Possible causes include differences in cell isolation or cell culture conditions or innate differences in the response of the cells to experimental circumstances.) (c) Our approach allows reliable identification of genes that are differentially expressed in a quantitative fashion without being completely absent in any lane. Such quantitatively regulated genes can be difficult to identify using the conventional single preparation approach because often a quantitative difference in band strength cannot be distinguished from loading differences between lanes, even when the strength on nonregulated bands is taken into account.
Although duplicate performance of experiments may seem time consuming, our experience has shown that the extra effort expended at this stage is more than repaid by reducing unnecessary cloning, sequencing, and Northern blotting later in the process.
Acknowledgments
This work was supported by grant no. Cu 31/2-1 to P.C. from the Deutsche Forschungsgemeinschaft. We are indebted to Prof. Karl Müller, Institut für Allgemeine Zoologie und Genetik, University of Münster, for his helpful guidance in this project and to Karin Tegelkamp for her excellent technical assistance.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  S. Brandt, S. Kloska, T. Altmann, and J. Kehr Using array hybridization to monitor gene expression at the single cell level J. Exp. Bot., December 1, 2002; 53(379): 2315 - 2323. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Sturtevant Applications of Differential-Display Reverse Transcription-PCR to Molecular Pathogenesis and Medical Mycology Clin. Microbiol. Rev., July 1, 2000; 13(3): 408 - 427. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
