Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 13 (4), 937-45

Experimentally Based Topology Models for E. Coli Inner Membrane Proteins


Experimentally Based Topology Models for E. Coli Inner Membrane Proteins

Mikaela Rapp et al. Protein Sci.


Membrane protein topology predictions can be markedly improved by the inclusion of even very limited experimental information. We have recently introduced an approach for the production of reliable topology models based on a combination of experimental determination of the location (cytoplasmic or periplasmic) of a protein's C terminus and topology prediction. Here, we show that determination of the location of a protein's C terminus, rather than some internal loop, is the best strategy for large-scale topology mapping studies. We further report experimentally based topology models for 31 Escherichia coli inner membrane proteins, using methodology suitable for genome-scale studies.


Figure 1.
Figure 1.
TMHMM topology prediction and probability profile for protein HISM_SALTY. The top line shows the predicted topology with the four predicted transmembrane helices. The thin black and gray curves show the posterior probabilities for inside and outside loop, respectively. The striped profile shows the probability for transmembrane helix. The lowest probability loop residue (LPLR) is indicated. The probability value for the LPLR (residue 108) is 0.43, reflecting the uncertainty in the topology predicted for this region. In the experimentally determined topology (Kerppola and Ames 1992), the LPLR is located in a transmembrane helix that is missed in the TMHMM prediction.
Figure 2.
Figure 2.
PhoA activities and GFP fluorescence intensities for the 34 inner membrane proteins. (A) Measured PhoA vs. GFP activities. The three encircled proteins (PotH, TdcC, YahN) have comparable PhoA and GFP activities and their C-terminal locations cannot be determined from the current data. (B) ln(PhoA/GFP) where PhoA is the measured PhoA activity normalized by the mean PhoA activity of all “active” PhoA fusions and GFP is the measured GFP activity normalized by the mean GFP activity of all “active” GFP fusions (see Results section). Because two different vectors were used for the GFP fusions (pWaldo and pGFPe), the GFP values were normalized by the mean fluorescence intensities of all “active” constructs for which the corresponding vector was used (the mean intensity for the pWaldo constructs is about threefold higher than for the pGFPe constructs). Black bars indicate proteins with a periplasmic C terminus; white bars, proteins with a cytoplasmic C terminus. The three gray bars correspond to the three encircled proteins in A for which the C-terminal location cannot be determined from the data.

Similar articles

See all similar articles

Cited by 33 articles

See all "Cited by" articles

Publication types

MeSH terms

LinkOut - more resources