Dear Drs. Limjindaporn and Sekaran,
It is true that prokaryotic expression of proteins evolved to function in eukaryotic hosts
is not an ideal setting. Nevertheless our previous experience with the bacterial twohybrid system (1,2) using several classes of protein baits, including heavily
glycosylated ones, proved to be very reliable when validated by co-ip in mammalian
expression systems. What is more likely is that protein domains with complex
secondary and tertiary structures fail to fold properly and putative interactions with
these domains are never realized during the screening. On the other hand more linear
domains and motifs will make excellent targets in any system. Every method of high
throughput screening has its disadvantages, being by forcing expression of membrane
or cytoplasmic proteins in the nucleus, or vice-versa (yeast two-hybrids) and the
reliability of these methods is estimated to be in the order of 50%. For this very reason
the development of complementary theoretical approaches for integrating data from
multiple experimental sources was put forward in order to improve the reliability of
protein interaction sets and facilitate biological validation based on predicted functions
assumed from topological analysis. Thus if one assumes that the data set is reliable
because all possible effort was made to control for false positives and negatives, and is
reproducible within the initial experimental setting (re-testing individual interacting
pairs), let the mathematical models point towards the best possible validation methods
to be used. We are now working on an RNAi –based assay knocking down specific
targets judged “testable” in an in-vitro virus infection-replication assay.
As to comply with the referees request we chose to perform a pull-down validation
assay for two interactions, one from cap and the other from Env, using HEK 293 cells
as an expression platform.
Methods
The same amplification reactions used in the cloning step of Env and cap into the pBT
bait vector (as described in the materials and methods section) were used to subclone
Env and cap into the pHM6 vector (Roche Applied Science, Indianapolis, IN) for an Nterminal in-frame fusion with an HA epitope tag. Full length PLG and CLU were
amplified by PCR using cDNA reversed transcribed from human peripheral blood
RNA, cloned into pCR4-TOPO (Invitrogen) and then subcloned in frame with GST
into the pET41a vector (EMD4Biosciences, Gibbstown, NJ). pET41 GST-PLG/CLU
was transformed into the E.coli expression host Rosetta DE3 (EMD4Biosciences) and
induced with IPTG from liquid cultures. Fusion protein from bacterial cell lysates was
purified with the aid of Glutathione-agarose columns (Thermo Scientific, Walthan, MA)
and purity verified by western blot with anti-GST antibody (Z5, Santa Cruz
Biotechnology, Santa Cruz, CA).
For pull-down assays HA-Env or HA-cap were transfected into HEK 293 cells using
Lipofectamine 2000 (Invitrogen) according to the manufacturer instructions. Forty-eight
hours after transfection cells were lysed in lysis buffer (50 mM Tris pH 7.6, 150 mM
NaCl, 1% Triton X-100, 1X protease inhibitor cocktail (EMD4Biosciences) at 4C for 30
minutes and cleared by centrifugation (21,000 g x 10 minutes). Supernatants were precleared by incubation with glutathione agarose (Santa Cruz Biotechnology) for 30
minutes at 4C under rotation. Two-hundred micrograms of each pre-cleared lysate was
incubated with 1 ug of either GST-CLU or GST-PLG for 3 hours at 4C under rotation.
Complexes were captured by glutathione-agarose beads, washed five times and eluted
by boiling in reducing Laemmli sample buffer. The material was resolved in a SDSPAGE, transferred to PVDF and immunoblotted with anti-HA antibody (Clone F-7,
Santa Cruz Biotechnology) followed by HRP-labeled secondary antibody. Blots were
revealed by ECL (GE Health Sciences, Pittsburgh, PA).
Results
Expression of Env produced a protein with an estimated mass of 57 KDa as detected by
the anti-HA antibody while cap appears as 12 kDa band with the same antibody. These
estimations are in line with previously published data for Env and cap expressed in
mammalian cell lines (3,4). GST-PLG was able to pull-down HA-Env from Envexpressing 293 cell lysates. No anti-HA signal was detected from pull-down of HA-capexpressing 293 cells, demonstrating the specificity of the interaction and pre-clearing
the 293-Env lysate with anti-HA antibody removed all the reactivity to the gel band
representing HA-Env. In a similar fashion we analyzed the ability of a GST-CLU fusion
protein to pull-down HA-cap from cap-expressing 293 cell lysates. GST-CLU
specifically pulled down HA-cap as no signal was detected from 293-Env cell lysates.
Although we did not attempt to validate any other interactions we hope that the
demonstration that dengue virus proteins expressed in a mammalian system can still
interact with mammalian proteins expressed in bacteria will be enough to satisfy the
referees doubts and also that of the reading public.
References:
1) Soares, LR et. al. 2004. Nat Immunol, 5:45-54
2) Lineberry N et.al. 2008. J Biol Chem, 283:28497-28505
3) Wei, HY et. al. 2003. Journal of Virological Methods, 109:17-23.
4) Limjindaporn T et. al. 2007. Biochem. Biophys. Res. Commun. 362:334-339.