Conservation of binding site

The web page on sequence logos on Wikipedia shows the sequence logo of the LexA-binding motif of Gram-positive bacteria. In this example we look at the other side: What is the amino acid sequence logo of the DNA-binding site of the LexA repressor? What is the consensus sequence?

We start by searching the NCBI Entrez database for lexA gene entries in the UniProtKB database and downloading them afterwards as GenPept file. In order to ensure that the file contains the desired entries, we check the entires for their definition (title) and source (species).

# Code source: Patrick Kunzmann
# License: BSD 3 clause

import matplotlib.pyplot as plt
import biotite.application.clustalo as clustalo
import biotite.database.entrez as entrez
import biotite.sequence as seq
import biotite.sequence.graphics as graphics
import biotite.sequence.io.genbank as gb

# Search for protein products of LexA gene in UniProtKB/Swiss-Prot database
query = entrez.SimpleQuery("lexA", "Gene Name") & entrez.SimpleQuery(
    "srcdb_swiss-prot", "Properties"
)
# Search for the first 200 hits
# More than 200 UIDs are not recommended for the EFetch service
# for a single fetch
uids = entrez.search(query, db_name="protein", number=200)
file = entrez.fetch_single_file(uids, None, db_name="protein", ret_type="gp")
# The file contains multiple concatenated GenPept files
# -> Usage of MultiFile
multi_file = gb.MultiFile.read(file)
# Separate MultiFile into single GenBankFile instances
files = [f for f in multi_file]
print("Definitions:")
for file in files[:20]:
    print(gb.get_definition(file))
print()
print("Sources:")
for file in files[:20]:
    print(gb.get_source(file))

Definitions:
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.
RecName: Full=LexA repressor.

Sources:
Verrucomicrobium spinosum DSM 4136 = JCM 18804
Opitutus terrae PB90-1
Fibrobacter succinogenes subsp. succinogenes S85
Mycobacterium tuberculosis H37Rv
Mycobacterium tuberculosis CDC1551
Clostridioides difficile 630 (Clostridium difficile 630)
Mycobacterium leprae TN
Mycobacterium tuberculosis H37Ra
Mycobacterium avium subsp. paratuberculosis K-10
Mycobacterium tuberculosis variant bovis BCG str. Tokyo 172
Mycobacterium tuberculosis variant bovis BCG str. Pasteur 1173P2
Mycobacterium tuberculosis variant bovis AF2122/97
Variovorax paradoxus S110
Tolumonas auensis DSM 9187
Teredinibacter turnerae T7901
Pectobacterium carotovorum subsp. carotovorum PC1
Listeria monocytogenes serotype 4b str. CLIP 80459
Geobacillus sp. WCH70
Geobacter sp. M21
Exiguobacterium sp. AT1b

The names of the sources are too long to be properly displayed later on. Therefore, we write a function that creates a proper abbreviation for a species name.

def abbreviate(species):
    # Remove possible brackets
    species = species.replace("[", "").replace("]", "")
    splitted_species = species.split()
    return "{:}. {:}".format(splitted_species[0][0], splitted_species[1])


print("Sources:")
all_sources = [abbreviate(gb.get_source(file)) for file in files]
for source in all_sources[:20]:
    print(source)

Sources:
V. spinosum
O. terrae
F. succinogenes
M. tuberculosis
M. tuberculosis
C. difficile
M. leprae
M. tuberculosis
M. avium
M. tuberculosis
M. tuberculosis
M. tuberculosis
V. paradoxus
T. auensis
T. turnerae
P. carotovorum
L. monocytogenes
G. sp.
G. sp.
E. sp.

Much better. For the alignment (required for sequence logo) we need to extract the slice of the sequence, that belongs to the DNA-binding site. Hence, we simply index the each sequence with the feature for the binding site and remove those sequences, that do not have a record specifying the required feature.

But we have still an issue: Some species seem to be overrepresented, as they show up multiple times. The reason for this is that some species, like M. tuberculosis, are represented by multiple strains with (almost) equal LexA sequences. To reduce this bias, we only want each species to occur only a single time. So we use a set to store the source name of sequences we already listed and ignore all further occurences of that source species.

# List of sequences
binding_sites = []
# List of source species
sources = []
# Set for ignoring already listed sources
listed_sources = set()
for file, source in zip(files, all_sources):
    if source in listed_sources:
        # Ignore already listed species
        continue
    bind_feature = None
    annot_seq = gb.get_annotated_sequence(file, include_only=["Site"], format="gp")
    # Find the feature for DNA-binding site
    for feature in annot_seq.annotation:
        # DNA binding site is a helix-turn-helix motif
        if (
            "site_type" in feature.qual
            and feature.qual["site_type"] == "DNA binding"
            and "H-T-H motif" in feature.qual["note"]
        ):
            bind_feature = feature
    if bind_feature is not None:
        # If the feature is found,
        # get the sequence slice that is defined by the feature...
        binding_sites.append(annot_seq[bind_feature])
        # ...and save the respective source species
        sources.append(source)
        listed_sources.add(source)
print("Binding sites:")
for site in binding_sites[:20]:
    print(site)

Binding sites:
VREIGNHFDISSTNGVRSILA
IREIGDAVGLTSTSSVAHQLR
VREICTAVGLRSTSTVHSHLN
IREIADAVGLTSTSSVAHQLR
IREIGDAVGLTSTSSVAHQLR
RAEIANELGFKSANAAEEHLQ
RAEIASELGFKSANAAEEHLK
RAEIAEILGFKSANAAEEHLK
RAEIAQQLGFRSPNAAEEHLK
VREIGEAVGLASSSTVHGHLA
VREIGEAVGLASSSTVHGHLA
VREIGEAVGLASSSTVHGHLD
RAEIAQRLGFRSPNAAEEHLK
RAEIAQRLGFRSPNAAEEHLK
VRELCDELGFKSPNTAHFHLK
VREICKAVGLSSTSSVHFHLK
RAEIARELGFKSPNAAEEHLK
RAEIARELGFRSANAAEEHLK
RAEIAKELGFRSANAAEEHLK
VREIARRFRITPRGAQLHLV

Now we can perform a multiple sequence alignment of the binding site sequences. Here we use Clustal Omega to perform this task. Since we have up to 200 sequences we visualize only a small portion of the alignment.

alignment = clustalo.ClustalOmegaApp.align(binding_sites)
fig = plt.figure(figsize=(4.5, 4.0))
ax = fig.add_subplot(111)
graphics.plot_alignment_similarity_based(
    ax, alignment[:, :20], labels=sources[:20], symbols_per_line=len(alignment)
)
# Source names in italic
ax.set_yticklabels(ax.get_yticklabels(), fontdict={"fontstyle": "italic"})
fig.tight_layout()

Finally we can generate our sequence logo and the consensus sequence.

profile = seq.SequenceProfile.from_alignment(alignment)

print("Consensus sequence:")
print(profile.to_consensus())

fig = plt.figure(figsize=(8.0, 3.0))
ax = fig.add_subplot(111)
graphics.plot_sequence_logo(ax, profile, scheme="flower")
ax.set_xticks([5, 10, 15, 20])
ax.set_xlabel("Residue position")
ax.set_ylabel("Bits")
# Only show left and bottom spine
ax.spines["right"].set_visible(False)
ax.spines["top"].set_visible(False)
fig.tight_layout()

plt.show()

Consensus sequence:
RREIADALGLRSPSAAEEHLK