Library Screening Weirdness

[schoenfeld at MSSCC.MED.UTAH.EDU]

Hi - 
I know this is a pretty standard protocol, but I've been having trouble
with my library screens.  I'm working with a lambda library which has small
(2-6 kb) genomic inserts.  Each time I screen, I get a handful of beautiful
positives from the high density (primary) screen. Since I do duplicate
lifts, I'm certain these aren't false positives.  When I core out these
positives, elute the phage & replate at 50-100 pfu/plate, EVERY plaque
lights up.  This has happened using 2 vastly different probes.

I've been suspicious about a variety of factors:
1.  my technique 
2.  the density of plaques - I've been working at 75,000 pfu/plate.  Is
this too thick?
3.  the membrane - I've heard anecdotal evidence in which a screen hasn't
worked when done on nylon filters but does work when nitrocellulose is
used.
4.  the hyb mix - I've been using 1 M NaCl, 1% SDS, 10% dextran and 100
ug/ml salmon sperm DNA.

to check the first 3 factors, I did the following expt.:
1.  created control plaques - I subcloned a 2 kb actin probe into lambdazap
arms and packaged.  Upon plating by blue/white screening, I picked and
amplified a blue plaque and a white plaque.  I then did phage rescue to
verify that the blue plaque was insertless and that the white plaque had my
actin fragment.

2.  "mixing" expt -
I plated out my genomic library with a few actin control plaques mixed in. 

plate 1 - 25000 pfu + 25 actin control plaques: lifted in duplicate using
nylon
plate 2 - 75000  "     "      "      "     "     "      "     "   "       "
        
plate 3 - 25000  "     "      "      "     "     "      "     "  
nitrocellulose  
plate 4 - 75000  "     "      "      "     "     "      "     "  
nitrocellulose
plate 5 - 25 blue plaques + 25 control actin plaques, lifted in duplicate
w/nylon
plate 6 - 25 blue plaques + 25 control actin plaques, lifted in duplicate
w/nitrocellulose

as an internal control, I also plated out the 25 control plaques (blue and
white) separately - this was to check that my titration was correct.  I
actually got 38 white and 36 blue, while the "mixed" plates (#5 and #6
above) had approx. 70 plaques each.

After nuking w/my actin probe (59 hyb, 55 wash in 3XSSC, as per published
protocol), I got some curious results.

My suspicions about choice of membrane & plaque density seemed unfounded -
I found very little difference between membranes and between plaque
density.  The number of double positives I got on each plate were in the
expected range.

What's curious is that on plates #5 and #6, the low density "mixed" plates,
again, every plaque lit up.  I have no idea what's going on.  The low
density filters were treated exactly the same as the high density filters,
from lifting through prehyb, hyb & washing.

Any ideas?

Thanks in advance
Robert