May 5, 1999, Paul E. Mead, PhD. Library construction.  Size selected (above 1.5 kb), double stranded cDNA was directionally cloned into pCDNA3 (InVitrogen, Ca) and transformed into E. coli TOP10F'.  The library consisted of approximately 3.3 x 106 independent clones (custom made by InVitrogen).  For screening, pools of approximately 100 independent clones were plate amplified, scrapped into liquid media and plasmid DNA was prepared (Wizard Prep, Promega, Wi). We have modified this to speed up the sib-selection and DNA isolation.  Approximately 100,000 cDNA clones from the library described above were arrayed in 384-well plates.  We use one of two companies for the arraying, Genome Systems or Research Genetics.  Pool size is something you will want to consider.  Low pool size increases sensitivity but increases number of pools you have to make. We screen pool s that each contain 96 cDNA clones.  Replicate the 384-well plates into 4 x 96-well plates.  Grow bacteria overnight at 37oC with gentle shaking.  Grow in 96-well plates as individual clones so there is no selection bias for large or poor growing clones.  Prepare DNA from entire plate, i.e. dump contents of plate into large weigh boat and transfer to 15 ml tube.  DNA  is prepared by Wizard prep (Promega, Wi). Vector considerations:  Check to ensure that the vector used to make the library will yield high levels of in vitro transcribed and translated proteins.  We use pCDNA3.0 from InVitrogen.  Not all vectors will give good protein yields, for example, pCDNA3.1 (in our hands) is not suitable for in vitro transcription and translation. In vitro transcription/translation of library pools and gel mobility shift assays.  Pools of protein were generated from cDNA library pools using TnT rabbit reticulocyte lysate transcription/translation kits according to manufacturers instructions (Promega, Wi).  Lysates were stored at -80oC until required.  One microgram of plasmid DNA for each pool was used for each transcription/translation reaction. Generally, due to cost, I do a 25 ml reaction rather than 50 ml.  This is enough for 5 gel shift reactions.  The best negative control  is probably an empty vector or some clone that you know will not bind the target site.  Unprogrammed lysates (i.e. no DNA added) can give higher background than you would see in a programmed lysate. Gel mobility shift assays were performed essentially as described (Crossley et al., 1995).  Briefly, gel shift reactions were prepared on ice as follows; a 2 to 5 ml aliquot of an protein pool is mixed with a gel shift cocktail containing (final concentration of a 20 ml reaction) 10 mM Hepes pH 7.8, 40 mM potassium glutamate, 5 mM magnesium chloride, 1 mM EDTA, 5% glycerol, 0.5 mM dithiothreitol, 50 mg/ml poly-dIdC (Sigma Chemical Co., Mo) and approximately 5 x 104 cpm 32P-labeled double stranded oligonucleotide probe.  After a twenty minute incubation on ice, gel mobility shifts were resolved on non-denaturing 5% (w/v) polyacrylamide gels in 0.5x TBE at 235 V for approximately 2 hours at 4oC.  Gels were transferred to Whatman 3M paper and dried.  Gel shift activities were visualized by exposing the gels to X-ray film overnight at room temperature. Ref:  Crossley, M., Merika, M. and Orkin, S. H.  Mol Cell Biol 15: 2448 (1995). For primary pools I always use 5 ml of TnT reaction.  If you have the luxury of a positive control that  you know binds to the site, it is a very good idea to clone this cDNA into the same vector that you will use for your library and try it out in the TnT gel shift experiment.  This can save you a lot of time in the future!   It is also important to determine the background that you can expect with a given probe.  Conditions ideal to the proteins that will bind need to be optimized so that background of the gel shift  is at a manageable level.  You can change a lot of parameters here.  Concentration of competitor (dIdC) is a good place to start.  Keep in mind that some background is a good thing .... gives an internal control (of sorts) for each gel shift reaction.  Note that the TnT products may shift differently than proteins extracted from cells.  Positive controls should be prepared at the same time, in the same vector as the library pools.  Try to get the probe to a reasonable size - long probes can cause problems.  I have used 50+ bp probes with success but the shifts can look pretty ugly - which does not matter as long as you can see the novel bands appearing.  If you have problems e-mail me. Heterodimeric partners:  If you suspect that the proteins that you are interested in may bind as heterodimers to the target site then the gel shift reaction can be supplemented with one of the heterodimeric partners.  This was very useful in our search for Mix family members as the hetrodimeric gel shift was often more robust (easier to see) than the  protein pool alone.  We added a truncated peptide encoding the DNA binding domain alone from Xenopus Mix.1 (HA:  Mead et al., Nature, 1996).  Each gel shift reaction was supplemented with approximately 1.0 ml of TnT reaction programmed with the HA plasmid. Sib-selection and characterization.  Single cDNA clones with specific gel shift activities were isolated from cDNA pools by sib-selection.  Briefly, the positive pools were replated on nitrocellulose (this is obviously not required if the library is arrayed into plates).  Individual clones were grown overnight in 96-well plates in LB broth.  DNA was prepared from rows of wells and the gel shift assay was repeated.  Rows of clones containing the original gel shift activity were subjected to the sib-selection procedure.  For this last step in the sib selection I grow 2 ml O/N cultures of each of the 12 clones from a positive row.   Prepare DNA and assay as above.  The 2 ml culture provides enough DNA to go directly on to sequencing and other analyses. The FROGS assay is rapid, taking only a few days to sib out positive clones.  One added bonus of this technique is that preliminary characterization can be performed prior to sib-selection.  For example, supershift and cold competitor assays can be performed on the primary pools to identify gel shifts resulting from known proteins.  This can save time by avoiding sibbing out known proteins.