Information Technology Reference
In-Depth Information
A
B
Figure 1. Prediction of signal peptides for gram-negative bacteria, using NN- and
HMM versions of SignalP. C, S and Y-scores represent, respectively, the output
from cleavage site networks, the output from signal peptide networks, and the output
of the combined cleavage site score, given by
. n-region: positively
charged amino-terminal segment; h-region: central hydrophobic segment; and c-
region: polar c-terminal segment. NrfA (SignalP-NN) - Positive results; Cut-off
between Ser28 and Thr29. NrfA (SignalP-HMM) - Positive results; Cut-off between
Gly23 and Cys24.
Y i
Ci
'
dSi
Both versions of Signal P predict that nrfH encodes for a signal peptide (based on y-
and S-score), with a maximum cleavage site probability between Ala29 and Met30 (Fig.2).
However, conflicting results were obtained when TMHMM 2.0 was run for the search of
transmembrane helices. Apparently, this subunit is expected to be a transmembrane protein,
with the bulk of the protein facing the periplasm. The N-terminus (residues1-6) remains in
the cytosol, while residue 7-29 is predicted to form a transmembrane helix, which most
likely acts as a membrane anchor (Fig.3). Similar profiles were obtained with JPRED - a
consensus method for protein secondary structure prediction - available at
http://www.expasy.org. Considering that a similar topological behaviour is often observed
among c-type cytocromes from bacteria [9 and references therein], these results suggest that
NrfA is devoided of a periplasm signal. Although not surprising for SignalP-NN, where the
discrimination between SAs and SPs has proved to be poor (according to S-score, 50% of
the SAs are predicted as SPs), critical reviews on the prediction of organellar targeting
signals [6] recommend the combination of SignalP with one of the available prediction
methods for transmembrane helices, as for example, PHDhtm and/or TMHMM. This
problem seems to be reasonably overstepped with the TMHMM program, which was
developed by E. L. Sonnhammer and co-workers [13] with an integrated architecture based
on SignalP-HMM and an HMM-based transmembrane helices prediction method. This
suggestion has revealed to be of particular usefulness with type II membrane proteins where
false positive results are often found.
Search WWH ::




Custom Search