Faculty of Health Sciences, University of Primorska, Slovenia; Phone: 00386 41 925 729; Email:
[email protected] Institute Jozef Stefan, Slovenia; Phone: 00386 41 253 526; Email:
[email protected]
New Methods and Techniques Poster Board Number: 776
DNA STRUCTURAL ALIGNMENT ALGORITHM CAN PREDICT PLASMID MOBILITY AND HOST RANGE BY LOCATING DNA SUBSTRATES FOR PLASMID TRANSFER Jan Zrimec
1,2
& Aleš Lapanje
1,3,4,
*
1
Faculty of Health Sciences, University of Primorska, Koper, Slovenia 2 Institute of Metagenomics and Microbial Technologies, Ljubljana, Slovenia 3 Saratov State University, Saratov, Russian Federation 4 Institute Jozef Stefan, Ljubljana, Slovenia
Introduction
Bacterial conjugation
Mobile element in circular form (plasmid)
Donor cell
Horizontal gene transfer (HGT) of conjugative plasmids that carry antimicrobial resistance (AMR) genes is a major health concern. The repertoire of bacterial hosts of such plasmids is characterized by their MOB group. MOB systems are defined by conjugation enzymes (relaxases) and DNA substrates (origin-of-transfer oriT regions). Based on the obervation that each oriT must have specific DNA conformational and physicochemical properties that most efficiently attract and enable activity of its particular relaxase, we: 1. Analyse the conservation of oriT structural properties among MOB groups, 2. Based on the findings develop algorithms for prediction of oriT regions and host repertoires in genomic DNA. 3. Implement webtools at http://dnatools.eu.
Integrative conjugative element
Origin-of-transfer (oriT) region
Initiation of conjugative transfer at oriT
Acceptor cell
Source: E Y Furuya & F D Lowy, Nat rev microbiol 4:36–45, 2006.
(National Institute of Health Image Bank)
Conserved oriT features inside MOB groups
Source: S Lang et al., Mol Microbiol 82(5):1071-1085, 2011.
Relaxase defined MOB groups
Conserved DNA structural properties
Mating pilus
Membrane
Relaxase
Auxilliary proteins
OriT region and relaxosome
DNA bubble
Structures improve discrimination of MOB groups To predict oriT structural properties including DNA stability, melting bubbles and curvature, parametric and dynamic DNA structure models were used (such as Nearest Neighbor and Peyrard-Bishop-Dauxois) and faster methods to predict melting bubbles were developed. Permutational analysis of variance was used to measure the amount of conservation of DNA structural properties and nucleotide seuqence in oriT regions (based on F-statistic and its p-value), which was F = 1.000, p < 0.001 and F = 0.525, P = 0.475, respectively.
Mobile DNA (plasmids and integrative conjugative elements) transfer from donor cells and integrate into genome of acceptor cells
oriT structural properties distinguish MOB groups
Method for fast prediction of DNA melting bubbles
J Zrimec & A Lapanje, IEEE Transact Comp Biol & Bionf 12(5):1137 - 1145, 2015.
Figure right. Amount of conservation of DNA structural properties vs. nucleotide seuqence at specific 10 bp interval positions inside oriT regions. The p-values of F-statistic are colored below the statistical significance level of 0.1 and below 0.05 are marked bright red to white. Using structural properties a higher amount of information is accessible in oriT.
Structure based algorithm predicts oriT locations and host repertoires STRA ST Using predicted oriT structural properties and machine Results obtained with structural alignment algorithm
... known regions in structural representation based on which a query is performed (100) … mobile elements structural representation (460)
Calculation of alignment p-value using bootstraping n_bins = 100
d_Bootstrap
Example of finding an oriT region in target by alignment to query
Based on the distance from the query seqeunce the most probable MOB group and host repertoires can be determined.
Distance from query sequence
query target
Frequency
learning algorithms, plasmid MOB groups were accurately sorted with a classification accuracy of 99.1%. However, to locate new oriT regions, dynamic homology searching – alignment algorithms (such as BLAST) are much more efficient than the machine learning approach. Therefore we developed a DNA structural representation and scoring metric suitable for use with alignment algorithms. Using this method we located oriT regions in over 80% of targets (p < 10-6).
Position in target sequence
Conclusions 1. Since results show that within particular MOB groups oriT regions have relatively conserved structural features, we propose that especially structural properties of oriT regions are connected to and coevolve with the proteins involved in DNA recognition, nicking and transfer reactions within their particular MOB group. 2. Based on merely 220 bp of DNA, our method can identify nic sites and transfer regions as well as most probable hosts and AMR pathways – implemented as webserver at http://dnatools.eu. 3. We hypothesize that different DNA sequences can form similar structural properties and are similarly efficient substrates for a specific type of relaxase -> By interfering with these properties oriT-relaxase interactions can be optimized or inhibited.