NotI passporting to identify species composition of complex microbial systems

Abstract

We describe here a new method for large-scale scanning of microbial genomes on a quantitative and qualitative basis. To achieve this aim we propose to create NotI passports: databases containing NotI tags. We demonstrated that these tags comprising 19 bp of sequence information could be successfully generated using DNA isolated from intestinal or fecal samples. Such NotI passports allow the discrimination between closely related bacterial species and even strains. This procedure for generating restriction site tagged sequences (RSTS) is called passporting and can be adapted to any other rare cutting restriction enzyme. A comparison of 1312 tags from available sequenced Escherichia coli genomes, generated with the NotI, PmeI and SbfI restriction enzymes, revealed only 219 tags that were not unique. None of these tags matched human or rodent sequences. Therefore the approach allows analysis of complex microbial mixtures such as in human gut and identification with high accuracy of a particular bacterial strain on a quantitative and qualitative basis.

INTRODUCTION

There is still much to learn about the human normal microflora. The human gut contains ∼1–2 kg of bacterial cells. The number of these cells in the intestine is 10–100 times larger than the number of cells in the human body but at best 10–15% of the microbial species are known, presumably because nobody can grow them in vitro in media due to the unknown growth requirements, oxygen sensitivity and other unidentified components (1). Many intestinal bacteria are known to provide molecules that the host itself cannot manufacture or degrade from nutritional compounds. Thus, these organisms are clearly of survival value, they are true symbionts of the host (2–5).

During the past 10 years enormous progress has been achieved in sequencing of bacterial genomes (6). Different approaches and techniques have been developed and now several bacterial genomes are being sequenced each month. However, despite impressive technological progress, mapping and sequencing of even small bacterial genomes is still expensive and laborious.

After completion of the genomic sequence from representative organisms, there is now a great demand for comparison with the genomes of other individuals, related species, etc. This growing field of comparative genomics is highly relevant for our understanding of human and animal health and disease, epidemiology, evolution and ecology (7–10). An area where this comparative technique can be rapidly and effectively applied is in the sequencing of related bacterial strains and species, in order to identify the genomic basis for differences in their biological properties, particularly pathogenicity, e.g. in the challenging task to study the human intestinal flora and to identify pathogenicity islands (9–12). The current strategies for mapping and sequencing are clearly not meeting the challenge of high-throughput comparative genomics. Alternative and complementary strategies need to be developed, and it is imperative now to find cost-effective and convenient methods that allow comparative genomics projects to be undertaken by a wide range of laboratories.

Recently we have developed a new efficient strategy for simultaneous genome mapping and sequencing. The approach is based on physically oriented, overlapping restriction fragment libraries called slalom libraries (13). Our model experiments have demonstrated the feasibility of the approach, and showed that the efficiency (cost-effectiveness and speed) of existing mapping/sequencing methods could be improved at least 5- to 10-fold. The benefits of the slalom approach will be most obvious in comparative sequencing experiments in combination with new high-throughput sequencing technologies which cannot be used in the shotgun sequencing approach. Even the short sequences (e.g. 20 bp) generated by pyrosequencing or massively parallel signature sequencing (MPSS) (14,15) should suffice, in principle. The combination of such novel sequencing techniques with the slalom approach would increase the power of the method 10- to 50-fold more. However, while the slalom approach can be used for the analysis of a particular bacterial species it cannot be used for analysis of complex microbial mixtures.

Identification of bacterial species and strains relies heavily on culture techniques. However, in a complex bacterial population, rapidly growing bacteria would overgrow, making quantification and identification of slow- or non-growing bacteria impossible. Techniques still have to be developed even to culture a representative selection of the microorganisms. Consequently, the picture of the intestinal flora has been biased in favor of the more easily cultured bacteria. There are some other methods available for biochemical analysis of complex microbial mixtures, e.g. by enzyme analysis which requires growth of colonies outside the body, or analysis of the composition of long fatty acids in stools which gives a crude indication of the composition of the normal flora. The limitations of such techniques are obvious (1,3,16).

The most popular approach that has so far been used to classify constituent strains in microbial mixtures without growing them in vitro is PCR amplification of the 16S rRNA genes. The amplified material can be used for fingerprinting or/and cloning and sequencing. However, both approaches are rather unspecific: different bacterial species can have the same sequence or PCR fragment and different PCR fragments/sequences can represent the same bacterial species. As 16S rRNA genes are conserved through evolution this approach cannot discriminate between closely related bacterial species or strains (12,16–19). Microarray-based techniques represent another alternative but they also have limitations (20–22). We have shown that the frequency profile of specific short oligonucleotides can be used to correctly classify bacterial genomes with an accuracy of 85%. However, at least 400 bases are needed for this approach (23).

Several directions of our previous and proposed studies were associated with human NotI jumping and linking clones (24–26). NotI linking clones contain DNA fragments flanking a single NotI recognition site, while NotI jumping clones contain sequences adjacent to neighboring sites. We have generated more than 22 000 unique human NotI flanking sequences (27). Analyses of these sequences have demonstrated that even short sequences surrounding NotI sites can yield important information allowing efficient isolation of new genes.

These results led to the realization that it would be feasible to develop a procedure for the analysis of complex microbial mixtures based on the short sequences surrounding NotI sites or, more general, on restriction site tagged sequences (RSTS).

MATERIALS AND METHODS

Growth of bacteria, other microbiology and molecular procedures were performed according to standard methods. Genomic DNA extraction from bacterial cells was performed according to the protocol at the http://www.uct.ac.za/depts/mmi/bbhelp/bac1.html. The protocol was based on the method of Marmur (1961) modified by J. L. Johnson (Virginia Tech, Blacksburg, VA).

The NotI-passporting procedure

Two oligonucleotides, BfocII: 5′-GGATGAAAACTGGA-3′ and Z98NOT: 3′-GTCGTGACTGGGAAAACCCTGGCCT ACTTTTGACCTCCGG-5′ were used to create the NotI linker.

Bacterial DNA (2 µg) at a concentration of 50 µg/ml was digested with 20 U NotI (Roche Molecular Biochemicals, Indianapolis, IN) at 37°C for 2 h, and then heat-inactivated for 20 min at 85°C.

Then, 0.4 µg of the digested DNA was ligated to NotI linker (50 M excess) overnight with T4 DNA ligase (Roche Molecular Biochemicals) in the appropriate buffer in 100 µl reaction mixtures. The DNA was then concentrated with ethanol and digested with 10 U BpmI at 37°C for 3 h.

Following digestion, BpmI was heat inactivated and the DNA was ligated overnight in the presence of a 50 M excess of the ZNBpm linker at room temperature. Two nucleotides: Zamine: 5′-CTCAAACCGT-3′ amine and Z2_univer: 3′- NNGAGTTTGGCACAGCACTGACCCTTTTGGGACC-5′ were used to create ZNBpm linker.

The sample was then purified using a JETquick PCR Purification Spin Kit (Genomed GmbH, Bad Oeynhausen, Germany), and dissolved in 100 µl TE. An aliquot of 1 µl of this sample was PCR amplified with Z1 univer (3′-GAG TTTGGCACAGCACTGACCCTTTTGGGACC-5′) and antiuniver (5′-CAGCACTGACCCTTTTGGGACC-3′) primers.

PCR was performed in 40 µl solution containing 67 mM Tris–HCl (pH 9.1), 16.6 mM (NH₄)₂SO₄, 2.0 mM MgCl₂, 0.1% Tween-20, 200 µM dNTPs, 3 µl PCR pool, 400 nM of each primer, and 5 U Taq DNA polymerase. The PCR cycling conditions were 95°C for 1.5 min, followed by 25 cycles of 95°C for 20 s, 66°C for 10 s, with 72°C for 10 s, with a final extension at 72°C for 1 min.

The final product was purified with the JETquick PCR Purification Spin Kit (Genomed) and cloned using TOPO TA Cloning kit (Invitrogen, Carlsbad, CA). Sequencing gels were run on ABI 377 automated sequencers (Applied Biosystems, Foster City, CA), according to the manufacturers’ protocols, using standard primers.

Sequence analysis

To facilitate RSTS analysis, restriction sites for all commercially available 8-bp restriction enzymes with adjacent 11 bp were extracted from the completed bacterial genome sequences available in EMBL database (release 71). In this way database comprising 19-nucleotide tags was prepared for subsequent analysis.

The analysis of sequences was performed at the Karolinska Institute Sequence Analysis Center (kisac.cgb.ki.se), using local versions of programs and public databases.

The database of nucleotide tags was compared with the draft human genome sequence and EMBL database (release 72) including all human and bacterial entries. Nucleotide similarity searches were performed with BLAST 2.2 (28,29) using the non-gapped alignment method (blast parameter g = F). The report cut-off (blast parameter -b) for the high-scoring segment pairs was set to 50. The statistical significance threshold (blast parameter -e) was set to: e = 1.E-1.

RESULTS AND DISCUSSION

Rationale for the experiments

The main idea was to develop new efficient methodologies for analysis of complex microbiological systems. As described above for the slalom approach, even short 20 bp sequence tags distributed over the whole genome suffice to generate a minimal contig of overlapping clones and therefore uniquely describe a bacterial genome.

We suggest use of only some specific fragments of the genomes (e.g. NotI linking clones, NotI tags, etc., refs 24,30) for analysis of the species composition in the intestinal microflora. Thus we do not aim to sequence complete genomes or study completely sequenced genes. We assign special signatures for the particular microorganism/genes and analyze these signatures in different samples of colonic flora. In this work we analyze the use of short sequence tags adjacent to NotI or other restriction enzyme recognition sites. The collection of NotI tags represents the NotI sequence passport or, in short, NotI passport. By the term ‘NotI passporting’ we mean the process of creating NotI tags/passports.

The naming is based on the enzymes originally used, but the methods can be adapted to other restriction enzymes as well.

The general design of the experiment is as follows (Fig. 1). DNA from fecal samples and surgical specimens is digested with NotI and subsequently a NotI passport for the particular specimen is generated. Comparing such passports from different individuals or from the same individual before and after drug treatment will reveal the differences between them. This information can in some cases be directly used to make conclusions. In other cases, using these sequences we can identify NotI linking clones that are different between two samples. These clones can be used for further analysis e.g. finding the genes that are involved in the examined matter (e.g. effects of antibiotic treatment, chronic bowel disease, etc.) or sequencing/isolation of the required microorganism.

Figure 1.

General scheme of the experiment. Two alternative approaches for the NotI passporting are shown: single tag sequencing and concatemer strategy. Two DNA samples are digested with NotI and ligated to a linker with NotI sticky ends. This linker contains the BpmI recognition sites. This restriction nuclease cut 16/14 bp outside of the recognition site. Then ligation mixtures are digested with this enzyme to generate 11/9 nucleotide tags adjacent to the NotI site. After several biochemical operations these tags are PCR amplified and ligated into concatemers that are cloned in plasmid vectors and sequenced (central part of the scheme). Alternatively, in single tag approach (shown at the left and the right) these PCR-amplified tags are directly cloned and sequenced. NotI tags from both DNA samples are collected and compared to identify the differences.

NotI passporting

We originally considered using the serial analysis of gene expression (SAGE) technique for this purpose. SAGE allows for both a representative and comprehensive differential gene expression profile (31). The idea of this approach is that, for each of the mRNA molecules, a short 10 bp sequence tag is produced (including recognition site for the tagging enzyme it is 14 bp) (32). Then these tags are ligated into concatemers and cloned. One sequencing reaction produces information for tens of RNA molecules. Thus by the sequencing of a few thousand clones one can expect to sample e.g. all of the estimated 10 000 to 50 000 expressed genes in a given cell population. Experiments with the SAGE technique for producing NotI tags were, however, unsuccessful. The complexity of genomic DNA in microbial mixtures is at least 100 times higher than the complexity of mRNA in eukaryotic cells. All RNA molecules must be tagged in SAGE but in our case, approximately one out of 250 molecules should be tagged. We propose to produce one tag for each 100–1000 kb, but in SAGE one tag is produced for every 256 bp. At the same time, a 14 bp tag does not suffice for unambiguous identification of sequences in genomic DNA. That is why we have developed a new procedure called Not passporting (see Materials and Methods). In this work we used the following modification (Fig. 2). Genomic DNA was digested with NotI and ligated to a linker with NotI sticky ends. This linker contained the BpmI recognition sites. This restriction nuclease cut 16/14 bp outside of the recognition site. The ligation mixture was digested with this enzyme to generate 11/9 nucleotide tags adjacent to the NotI site. This DNA sample was ligated to ZNBpm linker and PCR amplified with antiuniver and Z1univer primers to generate a 85 bp duplex. The final PCR-amplified molecule contained a 17 bp sequence tag which was missing 2 bp from the original NotI site (Fig. 2) and therefore the whole NotI tag contained 19 bp. NotI passports were experimentally produced for Escherichia coli K12, Enterobacter cloacae R4 and Klebsiella pneumoniae B4958. Experiments with samples obtained from mice demonstrated that the quality of DNA isolated from intestine or feces was sufficient to obtain NotI tags. The NotI passports uniquely identified these species and among 96 tags none was common for these three bacterial species. Of course, ditags or concatemers (31) also can be created from these 85 bp products after digestion with FokI or FokI + BpmI. We believe that new high-throughput technologies like MPSS will make sequencing of single tags a more efficient approach than creation of concatemers.

Figure 2.

Flow chart diagram explaining the generation of NotI tags. For producing ditags or concatemers the final PCR products may be digested with FokI or FokI + BpmI and ligated. BpmI recognition site is shown red and FokI in violet. The NotI tag sequence in the final PCR product stands out against blue background.

Restriction site tags for the analysis of complex microbial mixtures

As was pointed out above, this restriction site tagging procedure can be adapted to any recognition site for a restriction nuclease. It seems likely that for comprehensive analysis of flora composition use of several passports will be advantageous: different bacteria possess very different CG content. For NotI passports this means that bacteria having high CG content (NotI recognition site: GCGGCCGC) will be preferentially represented while using for example SwaI passports (SwaI: ATTTAAAT) will be better for bacterial genomes with high AT content. The choice of passporting strategy may also depend on the bacterial strains expected in connection with studies focused on particular problems such as cancer risk, medication, diet, etc.

We decided to test the potential of the passporting approach and analyzed 70 bacterial species that were completely sequenced (EMBL database, release 71). As an example, Table 1 shows the number of recognition sites for seven rare-cutting restriction enzymes in 30 bacterial species. It is easy to see that all 30 microbial species have different numbers of NotI recognition sites and therefore can be distinguished by NotI passporting. Moreover, from Table 1 we can see that the PmeI (GTTTAAAC) and SbfI (CCTGCAGG) restriction enzymes may be even more informative. Table 1 shows that different strains of E.coli and Helicobacter pylori have various numbers of NotI, PmeI and SbfI recognition sites. Thus all of these strains are uniquely defined by any of these enzymes and therefore this approach can really discriminate between different species and strains, something that is not possible with 16S rRNA genes sequencing (see also Fig. 3).

Table 1. The number of recognition sites for rare cutting restriction enzymes in selected bacterial species.

No.	Microbial genomes	Genome length (Mb)	Number of recognition sites
			NotI	PacI	PmeI	SbfI	SgfI	SgrAI	Sse2321
1	Aeropyrum pernix K1	1.7	88	8	29	62	5	135	48
2	Aquifex aeolicus	1.6	4	37	36	34	2	29	6
3	Bacillus subtilis	4.2	81	89	89	51	51	157	52
4	Borrelia burgdorferi	1.5	1	234	37	8	0	2	0
5	Buchnera sp. APS	0.7	0	184	12	5	2	1	0
6	Campylobacter jejuni	1.6	0	91	42	13	5	1	0
7	Chlamydophila pneumoniae AR39	1.2	2	59	10	21	13	4	1
8	Clostridium perfringens 13	3.1	1	479	51	6	0	2	0
9	Deinococcus radiodurans R1	3.3	15	1	4	28	7	645	164
10	Escherichia coli K12	4.6	23	143	87	68	222	548	31
11	Escherichia coli O157:H7	5.5	36	165	92	108	239	642	34
12	Helicobacter pylori 26695	1.7	7	32	35	4	88	61	12
13	Helicobacter pylori J99	1.6	14	34	43	4	87	66	15
14	Lactococcus lactis subsp. lactis	2.4	3	176	47	17	2	11	0
15	Mesorhizobium loti	7.6	1434	3	6	395	1041	4829	2686
16	Methanococcus jannaschii	1.7	0	258	40	8	0	7	4
17	Neisseria meningitidis MC58	2.3	74	24	27	28	11	360	176
18	Pasteurella multocida PM70	2.3	5	182	61	12	37	70	3
19	Pyrococcus abyssi	1.8	10	42	20	6	5	32	6
20	Pyrococcus furiosus DSM 3638	1.9	6	63	18	21	0	7	0
21	Rickettsia prowazekii	1.1	1	239	20	10	1	4	0
22	Salmonella typhi	4.8	89	129	79	132	276	768	238
23	Staphylococcus aureus Mu50	2.9	0	440	83	12	5	12	2
24	Streptococcus pneumoniae R6	2.0	1	40	25	30	1	9	0
25	Sulfolobus solfataricus	3.0	3	71	34	11	7	16	0
26	Synechocystis PCC6803	3.6	44	192	104	40	3160	182	18
27	Thermotoga maritima	1.9	10	4	5	14	33	117	7
28	Treponema pallidum	1.1	20	7	11	67	18	167	37
29	Vibrio cholerae	4.0	73	103	117	37	203	199	24
30	Yersinia pestis	4.7	65	207	87	50	111	447	22

Figure 3.

Analysis of specificity of RSTS for E.coli strains. Ditags and tags unique for E.coli species (A). Altogether 656 recognition sites (ditags) produce 1312 tags and among them 219 are non-unique. Analysis of strain-specific tags for E.coli K12 (B). Fifty-seven strain-specific restriction sites give 97 unique tags absent in other E.coli strains.

To further investigate the specificity of NotI, PmeI and SbfI tags, we compared these tags from three completely sequenced E.coli strains with all microbial and human sequences (EMBL database, release 72). The three E.coli strains contained altogether 1312 tags for these three enzymes and among them only 219 were not unique (Fig. 3A). The tag was considered unique if it matched only E.coli sequences. Importantly, since two tags describe the same restriction (e.g. NotI) site, one tag may be unique while the other can be non-unique and therefore both tags (unique and non-unique) may still represent a unique restriction site. In this particular case only 81 from 656 restriction sites were not unique. Figure 3B shows the results of a similar experiment but in this case only E.coli K12 tags were used for comparison with all other microbial sequences. Here the tag was considered unique if it matched only E.coli K12 sequences. Of course, the number (and proportion) of strain-specific tags was significantly less than the number (and proportion) of species-specific tags (97 unique versus 259 non-unique in Figure 3B against 1093 unique versus 219 non-unique in Figure 3A). In fact, all three E.coli strains had strain-specific tags. Even the O157H7 and O157H7 EDL933 strains could be distinguished and each had two unique tags.

These results demonstrate the power of the approach. In our comparative experiments we also analyzed EST and EMBL entries including all sequences. A vast majority of NotI tags were unique even if 1–2 nucleotide sequence mismatches were allowed.

As mentioned above a strong feature of NotI passporting is the internal controls. If a NotI site from a particular bacterial species yields, for example, NotItag 100 and NotItag 101 then we shall obtain both tags in approximately equimolar quantities, i.e. in similar numbers. If the situation is different (only NotItag 100 is present) then it most probably means that NotItag 100 originates from another (novel) bacterial species.

Table 2 shows the fraction of species-specific recognition sites (ditags) for NotI, PmeI and SbfI for 70 completely sequenced bacteria available in October 2002. We considered ditag as species specific if it matched sequences only from the same bacterial species. On average 96.6% of ditags was species specific and Listeria monocytogenes strain EGD had the lowest fraction of species-specific tags: 83.8%.

Table 2. Specificity of NotI, PmeI and SbfI recognition sites in 70 microbial genomes.

No.	Microbial genomes	Genome length (Mb)	Number of recognition sites				Species-specific unique restriction ditags (%)
			NotI	PmeI	SbfI	NotI + PmeI + SbfI
1	Aeropyrum pernix K1	1.7	88	29	62	179	100.0
2	Agrobacterium tumefaciens Cereon	5.7	609	9	258	876	96.9
3	Agrobacterium tumefaciens Dupont	5.7	608	9	258	875	96.9
4	Aquifex aeolicus	1.6	4	36	34	74	100.0
5	Archaeoglobus fulgidus	2.2	2	10	81	93	100.0
6	Bacillus halodurans C-125	4.2	23	109	22	154	100.0
7	Bacillus subtilis	4.2	81	89	51	221	98.2
8	Borrelia burgdorferi	1.5	1	37	8	46	97.8
9	Brucella melitensis strain 16M	3.3	144	7	53	204	87.7
10	Buchnera sp. APS	0.7	0	12	5	17	88.2
11	Campylobacter jejuni	1.6	0	42	13	55	100.0
12	Caulobacter crescentus	4.0	884	3	331	1218	99.6
13	Chlamydia muridarum	1.1	2	12	13	27	100.0
14	Chlamydia pneumoniae CWL029	1.2	2	10	21	33	100.0
15	Chlamydia trachomatis	1.0	3	13	18	34	100.0
16	Chlamydophila pneumoniae AR39	1.2	2	10	21	33	100.0
17	Chlamydophila pneumoniae J138	1.2	2	10	21	33	100.0
18	Clostridium acetobutylicum	4.1	0	62	16	78	100.0
19	Clostridium perfringens 13	3.1	1	51	6	58	100.0
20	Deinococcus radiodurans R1	3.3	15	4	28	47	95.7
21	Escherichia coli K12	4.6	23	87	68	178	88.2
22	Escherichia coli O157:H7	5.5	36	92	108	236	88.1
23	Escherichia coli O157:H7 EDL933	5.5	36	93	113	242	86.8
24	Haemophilus influenzae Rd	1.8	1	55	15	71	100.0
25	Halobacterium sp. NRC-1	2.6	791	6	81	878	92.8
26	Helicobacter pylori 26695	1.7	7	35	4	46	100.0
27	Helicobacter pylori J99	1.6	14	43	4	61	100.0
28	Lactococcus lactis subsp lactis	2.4	3	47	17	67	100.0
29	Listeria innocua Clip11262	3.0	8	69	2	79	84.8
30	Listeria monocytogenes strain EGD	2.9	8	52	8	68	83.8
31	Mesorhizobium loti	7.6	1434	6	395	1835	99.1
32	Methanobacterium thermoautotrophicum delta H	1.8	7	6	96	109	99.1
33	Methanococcus jannaschii	1.7	0	40	8	48	100.0
34	Mycobacterium leprae strain TN	3.3	251	9	83	343	98.5
35	Mycobacterium tuberculosis CDC1551	4.4	1214	2	163	1379	93.2
36	Mycobacterium tuberculosis H37Rv	4.4	1216	2	164	1382	93.4
37	Mycoplasma genitalium G37	0.6	0	57	0	57	100.0
38	Mycoplasma pneumoniae	0.8	2	62	1	65	98.5
39	Mycoplasma pulmonis	1.0	0	21	0	21	100.0
40	Neisseria meningitidis MC58	2.3	74	27	28	129	90.7
41	Neisseria meningitidis Z2491	2.2	76	26	30	132	92.4
42	Nostoc sp. PCC 7120	7.2	21	129	0	150	96.7
43	Pasteurella multocida PM70	2.3	5	61	12	78	100.0
44	Pseudomonas aeruginosa PA01	6.3	979	1	831	1811	99.5
45	Pyrobaculum aerophilum strain IM2	2.2	111	84	16	211	100.0
46	Pyrococcus abyssi	1.8	10	20	6	36	94.4
47	Pyrococcus furiosus DSM 3638	1.9	6	18	21	45	95.6
48	Pyrococcus horikoshii	1.7	11	23	9	43	95.3
49	Ralstonia solanacearum GMI1000	5.8	1375	1	572	1948	99.7
50	Rickettsia conorii Malish 7	1.3	2	35	11	48	93.8
51	Rickettsia prowazekii	1.1	1	20	10	31	90.3
52	Salmonella typhi	4.8	89	79	132	300	91.0
53	Salmonella typhimurium LT2	4.9	95	83	123	301	94.8
54	Sinorhizobium meliloti	3.7	760	3	215	978	90.3
55	Staphylococcus aureus Mu50	2.9	0	83	12	95	98.9
56	Staphylococcus aureus N315	2.8	0	81	13	94	100.0
57	Streptococcus pneumoniae R6	2.0	1	25	30	56	98.2
58	Streptococcus pneumoniae TIGR4	2.2	1	26	31	58	91.4
59	Streptococcus pyogenes M1 GAS	1.9	3	34	11	48	97.9
60	Sulfolobus solfataricus	3.0	3	34	11	48	100.0
61	Sulfolobus tokodaii	2.7	2	73	12	87	100.0
62	Synechocystis PCC6803	3.6	44	104	40	188	94.1
63	Thermoplasma acidophilum	1.6	9	3	43	55	100.0
64	Thermoplasma volcanium	1.6	3	20	25	48	100.0
65	Thermotoga maritima	1.9	10	5	14	29	100.0
66	Treponema pallidum	1.1	20	11	67	98	100.0
67	Ureaplasma urealyticum	0.8	0	25	1	26	100.0
68	Vibrio cholerae	5.1	73	117	37	227	98.2
69	Xylella fastidiosa	2.7	87	22	32	141	99.3
70	Yersinia pestis	4.7	65	87	50	202	91.1
	Total	199.7	11458	2708	5095	19261	96.6

To estimate efficacy of commercially available 8-bp cutters we measured the information value of these restriction enzymes for all 70 sequenced microbial genomes (EMBL database, release 72). If a particular genome contained no recognition sites for the analyzed restriction enzyme then this restriction enzyme got –0.5 points, if it contained 1–2 sites then this added 0 points, for 3–4 sites 0.5 points was added and so on, as follows: 5–105 sites: 1.0 point; 106–200 sites: 0.5 points; 201–500 sites: 0 points; more than 500 sites: –0.5 points. The results for seven enzymes in Table 1 are shown in Figure 4 and summary results in Table 3. As it was apparent from Table 1 and Figure 4, NotI, PmeI and SbfI are among the most informative enzymes.

Figure 4.

Schematic distribution of recognition sites for seven restriction enzymes in 70 completely sequenced bacterial genomes. Differently colored sections indicate different informative values based on the number of recognition sites in a particular genome (see text).

Table 3. Information value of recognition sites for rare cutting restriction enzymes in 70 microbial genomes.

Restriction enzyme	PmeI	SbfI	PacI	FspA	NotI	SgfI	SgrAI	SrfI	Sse2321	AscI	FseI	SwaI
Score	62.5	49.5	37	35.5	22.5	22.5	21.5	16.5	16	14	13	-21

Recently, a new DNA polymerase, Phi29 became commercially available (Amersham Biosciences, Buckinghamshire, UK). This DNA polymerase can be used for isothermal amplification of large DNA molecules including whole bacterial genomes (33,34). Sequence information obtained during passporting can be used for the design of PCR primers and for amplification of genomes from unknown bacterial species showing interesting features. This means that the RSTS approach gives a possibility both for studying biodiversity and for isolation of new microorganisms. It is also possible to use another enzyme instead of BpmI to obtain longer tags (G. Winberg and E. Zabarovsky, unpublished results).

The ability to analyze complex microbial mixtures is of great importance for many applications. For instance, differences between individual compositions of the normal flora will be instrumental for future analysis of the effects on the normal flora composition of diet, foods, geographical location, medication such as antibiotics and use of probiotics. Conversely, effects of intestinal microorganisms on colonic diseases, autoimmunity and colonic cancer risk can be evaluated. We suggest for this type of analysis to use the present approach, generation of RSTS. Hundreds of thousands of tags can be produced by a small research group in a short time allowing careful analysis of thousands of bacterial species/strains (31). We have demonstrated that such NotI tags can be efficiently produced and we have proven their high specificity. It is worthwhile to note that we have recently developed a new subtraction method (cloning of deleted sequences: CODE) and shown that this method can be efficiently applied to the NotI flanking sequences (30). Thus, the power and sensitivity of the passporting procedure can be significantly increased by removing the most abundant species with the CODE technique (35,36). We suggest to create a database for ‘NotI passports’ (as mentioned above, it is may be more correct to speak about ‘RSTS passports’). Such a database can be used together with a NotI (RST) microarray database (30) as these approaches are mutually complementary. This integrated database most probably will generate new knowledge as these two approaches are based on completely different biochemical techniques but aim to solve the same problem. As bacterial species are very different in nucleotide content we propose using combinations of different types of representation probes/microarrays (NotI, PmeI) for different applications.