iCR: a web tool to identify conserved targets of a regulatory protein across the multiple related prokaryotic species.

Gene regulatory circuits are often commonly shared between two closely related organisms. Our web tool iCR (identify Conserved target of a Regulon) makes use of this fact and identify conserved targets of a regulatory protein. iCR is a special refined extension of our previous tool PredictRegulon- that predicts genome wide, the potential binding sites and target operons of a regulatory protein in a single user selected genome. Like PredictRegulon, the iCR accepts known binding sites of a regulatory protein as ungapped multiple sequence alignment and provides the potential binding sites. However important differences are that the user can select more than one genome at a time and the output reports the genes that are common in two or more species. In order to achieve this, iCR makes use of Cluster of Orthologous Group (COG) indices for the genes. This tool analyses the upstream region of all user-selected prokaryote genome and gives the output based on conservation target orthologs. iCR also reports the Functional class codes based on COG classification for the encoded proteins of downstream genes which helps user understand the nature of the co-regulated genes at the result page itself. iCR is freely accessible at http://www.cdfd.org.in/icr/.

INTRODUCTION

Over last one and half decades, genomes of microorganisms have been sequenced at a highly accelerated pace. However, extracting useful information from such a large pool of genome data has become a major challenge of post genomics era. One approach to address this issue is to organize the large and complex genome into an ordered and manageable subsystem that can be tackled systematically. An important example of this approach is to study cellular processes and associated gene expression in terms of gene regulatory circuits. Each of these circuits contains a regulator and a list of its target sites (motifs) located upstream to a subset of genes that are being regulated (1–3). Such an approach will enable us to understand how the constituent genes of a genome come together to execute metabolic and physiological processes of a cell in response to a given regulator.

A large number of experimental and computational approaches are being attempted to understand how these genes come together to perform physiological function. The experimental approaches typically include microarray analysis of transcriptome (4,5). Subsequent to gathering the experimental data computational approaches are applied to search for common regulatory motifs and promoters present upstream to the up and down regulated genes and protein (6). Some of the computational tools like PHYLONET (7), BioProspector (8,9), Compare Prospector (9,10), MDscan (9,11), Motif Regressor (12), Bio Optimizer (13), PhyME (14) and so on are available for this purpose, but, most of these are either designed for eukaryotes or written to analyze the experimental data, such as micro array data, in terms of gene regulation.

An alternate approach could be to first select the regulator associated with a cellular process and then use computational approach to identify the potential target of regulatory protein which could then subsequently be followed up by experiments to validate the computationally identified targets. As a first step in this direction, we had previously proposed a tool called PredictRegulon, which finds targets of a regulatory protein in a genome based on limited set of known binding motif data (15). We have successfully used this tool to identify and validate the DtxR and IdeR targets in corynebacteria and mycobacteria, respectively (16,17). However an important limitation of Predictregulon was that it searches one genome at a time.

Carrying out simultaneous search in multiple genomes offers many advantages, most important among these are ability of such approach to reveal the conserved regulatory targets across the multiple related genomes. This would increase the confidence of experimental biologist in taking up experimental validation. Further it was also felt that if we could group the targets based on class of genes that is being regulated then we could provide the overall impact of the regulator on the physiology of the organism.

We describe here iCR (identify Conserved target of a Regulon), a web server tool, for identification of conserved high priority targets of a regulatory protein from heterologous sequence data of prokaryotes (which includes regulatory sequences of genes and their orthologs in other species) where the user can easily distinguish biologically important motifs from background noise based on their cross species conservation.

PROGRAM DESCRIPTION

iCR is a CGI based web application written in Perl and C language. It uses a Shannon relative entropy based profile search method, similar to what was used in PredictRegulon tool. This application can utilize the available experimental data on binding sites of a transcription regulatory protein (18–20) to identify the regulons of a given regulator in genomes of various phylogenetically related bacterial species.

iCR is composed of three parts (Figure 1): (i) a front-end web interface for submitting the block aligned known binding motifs and for selection of species of choice; (ii) a search engine for scanning the upstream sequences; and (iii) a classification and reporting system for rendering the textual output produced by iCR into a meaningful grouping. Each of these components is discussed in detail in the help pages linked to the iCR home page. A brief description is being given here.

Figure 1

Architecture of iCR. iCR is a CGI application which collects input from user using html forms (A). B represents a Perl script that gathers the input from A launches the Search Engine (C) which looks up genome sequences and their annotations (D), and returns the potential targets as an output which is further classified based on COG/Class or Genome. The classified output is returned as HTML output (F).

Input submission

iCR provides a web-based form for the input submission. The input form consists of two HTML pages. The first one accepts the sample motifs and the parameters defining the upstream region. On this page the known motifs can be copied either from sample input form or any authentic source and then be pasted in the web form in a block aligned fashion. The second page has a list of genomes organized in a taxonomically meaningful order for convenience in selection of multiple related species at a time and finally, the users need to specify the basis on which they want the predicted motifs to be grouped or classified on. The default or preferred option is Cluster of Orthologous Group (COG).

Search engine

Parameters accepted from the input forms are passed to a search engine which uses the Shannon relative entropy based profile scan method to scan the upstream sequences for regulatory motifs. This method is described in our previous paper PredictRegulon (15). However this analysis is carried out on multiple user selected genome and the results are compiled together. Since the complete COG data were not available for many of the genes of various genomes, we updated these data by running COGNITOR (21,22). Each COG selected represents the best hits to proteins from at least three lineages.

The output of the search result is classified and grouped based on one of the three options—orthology, function class code or genome. Classification based on orthology (default option) lists all the orthologous targets of a regulator together emphasizing the fact that these are conserved targets of a given regulon.

Output

All the predicted and classified target motifs are presented as HTML table. This table has following columns: COG name, Functional class code, Genome, motif score, motif, Gene id mentioned in NCBI's ptt table, ORF number and gene product. The program predicts a number of motifs, the blue background color shows the high scoring motifs above the cut-off value. The motifs with yellow background color depicts exact match to the known binding sites.

Example usage

To demonstrate the typical application of iCR's regulon assignments, we chose to use known LexA-binding sites from Bacillus subtilis as a query set. These sites were collected from PRODORIC (19). We then selected different species belonging to Fermicutes (Bacillales, Lactobacillales, Clostridia and Mollicutes) simultaneously for search. We obtained the result classified on COG in which DNA motifs upstream to lexA (COG1974), recA(COG0468), uvrB(COG0556), dinP(COG0389), rpsE(COG0098), rpsN(COG0098), rggD (COG0457) and so on were picked up in many species together and therefore they qualify for conserved targets of LexA regulon (Table 1). Lex A is known to autoregulates itself (23). recA gene has been experimentally shown to be part of LexA regulon in Escherichia coli as well as B.subtilis (23,24). Homologs of dinP have also been shown to be regulated by LexA protein in Bdellovibrio bacteriovorus (25). LexA protein has been reported to interact with the regulatory region of uvrB in B.subtilis (19). All these observations confirm that the program is capable of identifying significant and high priority targets of a given regulator successfully. Additionally the result also highlights many motifs upstream to hypothetical genes/ORFs. An experimental confirmation of interaction of these motifs to LexA, followed by a functional assay based on known processes involved with a given regulator, could shed more lights on function of these hypothetical genes.

Table 1

Output of iCR showing the conserved targets of LexA regulon in Fermicutes

COG	Class	Genome	Score	Position	Site	Gene	Synonym
COG1974	K	NC_004193	4.6875	−77	AGAACGAGTGTTTG	lexA	OB1669
COG1974	K	NC_003030	4.77125	−84	AGAACATAAGTTTG	lexA	CAC1832
COG1974	K	NC_002745	4.88271	−71	CGAACAAATGTTTG	lexA	SA1174
COG1974	K	NC_004557	4.82946	−80	AGAACATAAGTTTG	lexA	CTC01298
COG1974	K	NC_003366	4.83493	−70	AGAACATAAGTTTG	lexA	CPE1161
COG1974	K	NC_002570	4.72756	−77	AGAACTTATGTTTG	lexA	BH2356
COG1974	K	NC_000964	4.81601	−118	CGAACCTATGTTTG	lexA	BSU17850
COG1974	K	NC_003923	4.88303	−71	CGAACAAATGTTTG	lexA	MW1226
COG1974	K	NC_003212	4.82162	−79	CGAACCTTTGTTTG	—	LIN1340
COG1974	K	NC_002758	4.88182	−138	CGAACAAATGTTTG	lexA	SAV1339
COG1974	K	NC_003210	4.81541	−79	CGAACCTTTGTTTG	—	LMO1302
COG0468	L	NC_002570	4.64423	−121	CGAATAAATGTTCG	recA	BH2383
COG0468	L	NC_003212	4.67474	−138	CGAATAAATGTTCG	recA	LIN1435
COG0468	L	NC_003210	4.66915	−138	CGAATAAATGTTCG	recA	LMO1398
COG0468	L	NC_003923	4.40442	−143	AGCACGTTTGTTCG	recA	MW1168
COG0468	L	NC_002758	4.40302	−80	AGCACGTTTGTTCG	recA	SAV1285
COG0468	L	NC_003030	4.90549	−48	AGAACAAATGTTCG	recA	CAC1815
COG0468	L	NC_003366	5.01207	−34	AGAACTTATGTTCG	recA	CPE1673
COG0468	L	NC_004461	4.42484	−143	AGTACGTTTGTTCG	—	SE0963
COG0468	L	NC_000908	4.18494	−236	TGAACTGTTGTATG	recA	MG339
COG0468	L	NC_002745	4.40405	−143	AGCACGTTTGTTCG	recA	SA1128
COG0468	L	NC_004557	4.9426	−54	AGAACAGATGTTCG	recA	CTC01289
COG0556	L	NC_000964	4.7767	−122	CGAACTTTAGTTCG	uvrB	BSU35170
COG0556	L	NC_003923	4.8228	−105	CGAACAAACGTTTG	uvrB	MW0720
COG0556	L	NC_002745	4.82248	−105	CGAACAAACGTTTG	uvrB	SA0713
COG0556	L	NC_003030	4.93204	−29	CGAACAAATGTTTG	uvrB	CAC0502
COG0556	L	NC_002758	4.82157	−103	CGAACAAACGTTTG	uvrB	SAV0758
COG0556	L	NC_004193	4.65391	−69	CGAATACTTGTTCG	—	OB2488
COG0556	L	NC_003212	4.62091	−158	CGAAAATATGTTCG	uvrB	LIN2632
COG0556	L	NC_003210	4.61721	−160	CGAAAATATGTTCG	uvrB	LMO2489
COG0556	L	NC_004461	4.90087	−128	CGAACAAATGTTTG	—	SE0541
COG0389	L	NC_003366	4.82409	−26	TGAACATATGTTTG	dinP	CPE1566
COG0389	L	NC_003923	4.77999	−49	GGAACACGTGTTCG	—	MW1251
COG0389	L	NC_002758	4.33641	−6	AGAACATTTGTTCT	—	SAV1364
COG0389	L	NC_002745	4.81919	−49	AGAACACGTGTTCG	—	SA1196
COG0389	L	NC_003210	4.72978	−33	AGAACGCTTGTTCG	—	LMO1975
COG0389	L	NC_004461	4.32424	−75	AGAACAAATGTTCT	—	SE1046
COG0389	L	NC_003212	4.73647	−33	AGAACGCTTGTTCG	—	LIN2082
COG0389	L	NC_004557	4.82946	−40	AGAACATAAGTTTG	—	CTC00437
COG0389	L	NC_000964	4.37402	−68	CGAACATAAGTTCT	yqjW	BSU23710
COG0199	J	NC_004368	4.29015	−280	TGAACGTATGTACG	—	GBS0071
COG0199	J	NC_002662	4.9713	−280	CGAACGTATGTTCG	rpsN	L0391
COG0199	J	NC_003028	4.22998	−280	TGAACGTATGTACG	—	SP0222
COG0199	J	NC_003098	4.22977	−280	TGAACGTATGTACG	rpsN	SPR0202
COG0199	J	NC_002737	4.41534	−278	CGAACGTATGTACG	rpsN	SPY0064
COG0199	J	NC_003485	4.41477	−278	CGAACGTATGTACG	rpsN	SPYM18_0065
COG0199	J	NC_004432	4.29794	−140	CGAAATTGTGTATG	—	MYPE10040
COG0199	J	NC_004070	4.41397	−278	CGAACGTATGTACG	rpsN.1	SPYM3_0053
COG0199	J	NC_004116	4.2898	−280	TGAACGTATGTACG	—	SAG0071
COG1396	K	NC_003485	4.21446	−8	AGAAACCATGTTAG	—	SPYM18_0038
COG1396	K	NC_003923	4.32136	−263	GGAACAAGTGTACG	—	MW1228
COG1396	K	NC_004070	4.21434	−8	AGAAACCATGTTAG	—	SPYM3_0031
COG1396	K	NC_002570	4.4873	−118	GGAACGGGCGTTTG	—	BH0096
COG1396	K	NC_003028	4.47861	−127	TGAACAAATGTTGG	—	SP1115
COG1396	K	NC_002737	4.21453	−8	AGAAACCATGTTAG	—	SPY0037
COG1396	K	NC_004193	4.36271	−253	TGAACAGGAGTTAG	—	OB3501
COG1396	K	NC_003366	4.35319	−58	TGAACATTTGATTG	—	CPE2564
COG0098	J	NC_003028	4.38376	−109	AGAAGTGGTGTTCG	—	SP0227
COG0098	J	NC_004116	4.25066	−110	TGAAGTGGTGTTTG	rpsE	SAG0075
COG0098	J	NC_002737	4.23373	−110	TGAAGTGGTGTTTG	rpsE	SPY0069
COG0098	J	NC_004368	4.25082	−110	TGAAGTGGTGTTTG	rpsE	GBS0075
COG0098	J	NC_003098	4.38367	−109	AGAAGTGGTGTTCG	rpsE	SPR0206
COG0098	J	NC_004070	4.23345	−110	TGAAGTGGTGTTTG	rpsE	SPYM3_0057
COG0098	J	NC_004350	4.24208	−109	TGAAGTGGTGTTTG	rs5	SMU.2009
COG0098	J	NC_003485	4.23361	−110	TGAAGTGGTGTTTG	rpsE	SPYM18_0069
COG0457	R	NC_004557	4.34686	−240	GGAAGAAGAGTTTG	—	CTC02554
COG0457	R	NC_002570	4.38934	−268	CGAAGCAACGTTTG	—	BH3054
COG0457	R	NC_004557	4.39536	−233	AGAACAATTGTATG	—	CTC01089
COG0457	R	NC_002745	4.37946	−17	AGAAATGAGGTTCG	—	SA1448
COG0457	R	NC_003923	4.3797	−17	AGAAATGAGGTTCG	—	MW1570
COG0457	R	NC_003098	4.47855	−97	TGAACAAATGTTGG	rggD	SPR1022
COG0457	R	NC_002758	4.3788	−86	AGAAATGAGGTTCG	—	SAV1620

Note: Gene, Synonym column is as per NCBI ptt table. Class codes—K involved in transcription, L in DNA replication, recombination and repair, J represents orthologs involved in translation, ribosomal structure and biogenesis and so on.

To test the sensitivity of the iCR predictions, we deleted two important and known binding motifs of LexA protein (present upstream to the dinB and uvrB in B.subtilis) from the input form and selected two species of Bacillales, B.subtilis and Bacillus holodurans. These two motifs were picked up on result page with blue background proving the reliability of predictions.

Certainly iCR results can serve as a useful starting point for molecular and cellular biologists for designing experiments to see the in vitro and in vivo effects of a regulatory protein in different systems.

CONCLUSION

To summarize, iCR is a web server that permits high throughput, detailed and fully automated prediction of potential binding targets of a regulatory protein in user selected prokaryotic species. iCR consists of 115 prokaryotic species arranged phylogenetically on the web interface. The first column on the result page, COG, is hyperlinked to NCBI and are fully navigable to allow users to have easy access to more related and descriptive information. The genome column shows the genome ID that is hyperlinked to a HTML page containing genome names corresponding to different IDs. For the user's convenience, functional class code column has also been linked to a page, which has a description of all the codes. iCR's strengths are in its free web accessibility, its comprehensiveness regarding choice of multiple species at a time, sorting of result based on COG and Class, and its interactive graphical interface.