variability in Liquid b-gal

David Stillman stillman at
Thu Sep 19 10:13:51 EST 1996

In article <kalch-1709961154130001 at> Michael Kalchman,
kalch at writes:
>Subject: variability in Liquid b-gal
>From: Michael Kalchman, kalch at
>Date: Tue, 17 Sep 1996 11:54:13 -0800
>does anybody know the varibility in the liquid beta-gal assay?  I seem to
>have quite a large standard deviation when a large number of assays are
>done on a similar sample.
>Any insight and references are greatly appreciated.

We have had a LOT of experience with assays for lacZ activity in extracts
using ONPG as a substrate, and we have found a number of variables which
can affect the standard deviation. 

First, integrated reporters always give much higher reproducibility than
do plasmid numbers. 

Many people measure the OD of the cells, prepare an extract for the ONPG
measurement, and then normalize the ONPG value to the cell number of
cells (OD) to get a lacZ value. We use a different method. We make a
lysate by vortexing with glass beads. One aliquot of the supernatant is
assayed for lacZ activity by incubating with ONPG in the standard way,
and a second aliquot is used to measure protein levels by Bradford assay.
The ONPG value is normalized to the protein level to get a final lacZ
number. Typically there is 200ul of lysate, and we use 5 - 50 ul for the
ONPG measurement and 2 - 20 for the Bradford; thus there is extract left
to repeat either measurement with more or less extract. 

We get much better reproducibility with the assay normalizing lacZ to
protein level instead of normalizing to OD of cells. I suspect that after
one measures the OD of cells, there can be variability in recovery of
cells and in the lysis, leading to wider standard deviations. With the
Bradford method, one is normalizing to the exact same extract. 

There can be real problems with variability in the data if one does not
limit the values to the linear range of the assay. For the ONPG assay one
measures the OD at 420 nm. We only accept values in the linear range,
between 0.1 and 0.35. Similarly, the linear range for the Bradford is OD
595 between 0.1 and 0.6. 

We use an Excel macro for our calculations. We enter the ONPG values (OD
420), the Bradford values (OD595), the volumes used, and the times of the
reactions (min.) into the spreadsheet, and it calculates the lacZ value
along with standard deviations for replicates. E-mail me if anyone would
like a copy of this macro. 

Like Susan Forsburg (forsburg at, we always assay at least three
independent transformants grown in separate liquid cultures. 

The ONPG assay has a time component, with the length of time of the assay
is factored into the final value. I agree with Marieke Koedood
(rkoedood at that very short incubation times can lead to poor
data, as it is difficult to precisely determine the reaction time. We get
the best data for incubations between 20 and 120 min. 

Another cause of error is variability in pipetting. A sample with high
lacZ activity might turn yellow very quickly, and so one might use only 1
or 2 ul for the assay. Now the problem is that the most pipettors are
quite variable at this volume, leading to wide standard deviations. One
is advised to dilute the sample and add a larger volume. 

Finally, one observation about experimental technique with lacZ assays.
There have been many new students who have performed lacZ experiments in
my lab, and their numbers are always terrible on their first couple of
tries with this assay. For reasons that I don't understand, they always
get better at the assay, and their standard deviations improve. I wish I
knew what they are doing differently as they work with the assay, but it
always gets better! 

David Stillman
Division of Molecular Biology and Genetics
Department of Oncological Sciences
University of Utah Health Sciences Center 
Salt Lake City, UT  84132
stillman at
phone: 801-581-5429
fax: 801-581-3607

More information about the Yeast mailing list