Sequence Analysis of a 101-Kilobase Plasmid Required for Agar Degradation by a Microscilla Isolate

Abstract

An agar-degrading marine bacterium identified as a Microscilla species was isolated from coastal California marine sediment. This organism harbored a single 101-kb circular DNA plasmid designated pSD15. The complete nucleotide sequence of pSD15 was obtained, and sequence analysis indicated a number of genes putatively encoding a variety of enzymes involved in polysaccharide utilization. The most striking feature was the occurrence of five putative agarase genes. Loss of the plasmid, which occurred at a surprisingly high frequency, was associated with loss of agarase activity, supporting the sequence analysis results.

Bacteria have long been exploited as a source of natural products for use in medicine, agriculture, and industry. Until recently, most studies utilized microbes that had been isolated from clinical or terrestrial environments. It is now recognized that bacteria living in marine environments provide an abundant and as-yet-untapped source of metabolic properties (8). Marine sediments are rich in bacteria, with reports of 10⁸ to 10¹⁰ microbes per g of sediment for the first few centimeters of sediment (21). Estimates of bacterial diversity in natural environments have indicated that, while a few organisms may predominate, such environments still represent a highly complex assemblage of microbes (45).

We have initiated a study to identify and characterize the genes present on plasmids isolated from bacteria present in coastal marine sediments. Extrachromosomal elements by definition encode functions that are not essential for cell growth but which provide an advantage to the host bacterium under certain growth conditions. It is therefore not surprising that a wide variety of traits in bacteria have been found to be plasmid encoded. In an earlier study (41), ca. 30% of more than 1,000 aerobic heterotrophic bacteria isolated from coastal California marine sediments contained at least one plasmid that ranged in size from 5 to >250 kb. These plasmids appeared to contain novel and generally uncharacterized replication regions since no homology was detected between ca. 300 plasmids of marine origin (41) and 15 replicon probes derived from plasmids found in bacteria isolated from mammalian or terrestrial sources (10). These findings suggested that plasmids in marine sediment microbial communities are a unique and diversified set of extrachromosomal elements.

We present here the characterization of an agar-degrading marine isolate, Microscilla sp. strain PRE1. This organism contains a 101-kb plasmid, designated pSD15, which potentially encodes five different agarases and is essential for the ability of the bacterium to degrade agar. The complete DNA sequence and analysis of pSD15 are reported.

MATERIALS AND METHODS

Isolation and characterization of marine bacterial isolate PRE1.

Marine sediment associated with the roots of pickleweed (Salicornia virginica) was collected from the Kendall-Frost Mission Bay Marsh Reserve in San Diego, Calif. The sediment was resuspended in artificial seawater (0.3 M NaCl, 0.1 M KCl, 0.01 M CaCl₂, 0.05 M MgSO₄), and serial dilutions were plated on M7 medium (0.1% tryptone in artificial seawater) solidified with 15 g of agar per liter. The plates were incubated for 2 to 4 days at 30^°C. Isolated colonies, selected on the basis of differing morphology and color, were patched on to M7 medium and the bacteria were screened for total plasmid content by a modification of the Kieser method (22) as described previously (41). One isolate, PRE1, which appeared to degrade agar as detected by pitting around each colony and which contained a single covalently closed circular DNA plasmid ca. 100 kb in size, was selected for further study.

The morphology and cell size of isolate PRE1 were determined in an exponentially growing culture by measuring 100 cells after staining of the cell membrane with FM4-64 at 5 μg/ml (Molecular Probes, Eugene, Oreg.) and of the DNA with DAPI (4′,6′-diamidino-2-phenylindole) at 5 μg/ml. Cells were also Gram stained by the standard method. Growth at 25, 30, 37, 42, and 45°C was monitored in M10 broth (0.25% yeast extract–0.4% tryptone, prepared in artificial seawater). The ability to grow in different NaCl concentrations was determined by shaking cultures for 24 h at 30^°C in M10 broth modified to contain NaCl concentrations from 0.5 to 6%. The growth requirement for K⁺, Mg²⁺, or Ca²⁺ was tested in liquid M10 medium prepared with artificial seawater missing the cation in question.

Sensitivity to antibiotics was tested for both isolate PRE1 and a derivative cured of plasmid pSD15. Antibiotic discs (Difco/BD Biosciences, Franklin Lakes, N.J.) containing gentamicin (10 μg), kanamycin (10 μg), neomycin (30 μg), streptomycin (10 μg), ampicillin (10 μg), trimethoprim (5 μg), rifampin (5 μg), tetracycline (30 μg), erythromycin (15 μg), carbenicillin (100 μg), chloramphenicol (30 μg), nalidixic acid (30 μg), and novobiocin (30 μg) were applied to lawns of the two strains spread on M10 plates. Plates were incubated for 24 to 36 h at 30°C, and resistance was determined by the method of Bauer et al. (5).

Identification of PRE1 by using 16S ribosomal DNA (rDNA) sequence.

For 16S rDNA analysis, genomic DNA was purified from 1 ml of an overnight culture of PRE1 grown at 30°C in M10 broth by using the QIAampTissue Kit (Qiagen, Chatsworth, Calif.). The 16S rDNA gene was amplified from 0.1 to 0.5 μg of genomic DNA by using the primers fD1 and rD1 (47) and Taq2000 DNA polymerase (Stratagene, La Jolla, Calif.) as described previously (42). The amplified products were electrophoresed on a 0.8% agarose gel, and a fragment of ca. 1.5 kb was purified by using GeneClean II (Q-Biogene, Carlsbad, Calif.). Partial DNA sequences were obtained for the 1.5-kb fragment by using three primers corresponding to the following positions of the Escherichia coli 16S rRNA gene sequence: primer A, positions 519 to 536; primer B, positions 907 to 926; and primer C, positions 1392 to 1406 (25). Sequence alignment was carried out by using the Ribosomal Database Project II online analysis tools with the SSU prokaryotic data set (26).

Plasmid pSD15 stability.

A single colony of Microscilla sp. strain PRE1 able to degrade agar (as evidenced by pitting of the solid medium) was picked from a fresh plate and resuspended in 5 ml of M10 broth. The culture was incubated at 30°C with shaking, and at 15 h (mid-log phase), 20 h (late log phase), and 26 h (stationary phase), a 10-μl aliquot was removed, serially diluted, and spread on M10 agar plates. After incubation at 30°C for ca. 36 h, by which time tiny colonies had appeared, single colonies were patched onto M10 plates, which were then incubated at 30°C. The number of patched colonies capable of degrading agar could be determined by direct visualization after 18 to 24 h. The assay was repeated twice, starting each time with a unique single colony isolate. The results reported are for a total of 500 colonies for each growth phase from the three repetitions.

Isolation of supercoiled plasmid DNA.

A 5-ml overnight culture of Microscilla sp. strain PRE1 grown in M10 broth was transferred to 1 liter of M10 broth and incubated with vigorous aeration at 30°C. To avoid excess polysaccharide formation, the culture was harvested after ca. 13 h of incubation (mid-exponential phase). Supercoiled plasmid DNA was prepared as described previously (42) by using the alkaline lysis method of Birnboim and Doly (9), except that RNase A was omitted from the first solution and that the sample was not extracted with phenol-chloroform prior to precipitation by isopropanol. Plasmid DNA was subsequently purified by two rounds of cesium chloride-ethidium bromide gradient centrifugation.

Plasmid library construction.

A small-insert random-fragment library was constructed from purified pSD15 DNA as follows. First, 5 to 10 μg of pSD15 DNA was sheared by sonication, treated with Bal 31 nuclease followed by T4 DNA polymerase to repair the ends, and then size fractionated by agarose gel electrophoresis. Fragments 2 to 4 kb in size were recovered from the gel and ligated into the SmaI site of commercially available bacterial alkaline phosphatase (BAP)-treated pUC18 (Amersham Pharmacia Biotech, Piscataway, N.J.). The ligation mixture was subjected to StrataClean Resin (Stratagene, La Jolla, Calif.) treatment to remove ligase and then transformed into E. coli XL1-Blue by electroporation. An aliquot of the transformants was plated on a Luria-Bertani (LB) agar containing ampicillin (100 μg/ml), X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; 32 μg/ml), and IPTG (isopropyl-β-d-thiogalactopyranoside; 32 μg/ml) to determine the percentage of recombinants based on blue and white screening. The range of insert sizes was determined by a convenient whole-cell cracking procedure (Promega Protocols and Application Guide, 2nd ed., 1991 [Promega, Madison, Wis.]).

DNA sequencing and data annotation.

Sanger sequencing reactions were performed on 1,824 randomly chosen subclones by using BigDye Terminators (Applied Biosystems) and then analyzed on ABI377 automated sequencing machines (Applied Biosystems). This generated seven- to eightfold sequence coverage across the 100-kb plasmid. The sequences were evaluated for quality and error probability by using the program phred (15), assembled by using the phrap assembler (14), and viewed by using consed (16). All contigs were ordered and oriented, and all gaps were closed by using a directed primer walking strategy. A final quality value of phred40 (1-base error in 10,000 bases) with no gap regions, double coverage, or two chemistries across single-stranded areas was achieved.

The pSD15 DNA sequence was analyzed by using the computer programs Vector NTI (version 5.5; InforMax, North Bethesda, Md.) and the GCG Wisconsin Package (Oxford Molecular Group). The initial assignments of putative open reading frames (ORFs) were made according to the criteria that (i) the start codon was ATG or GTG; (ii) the stop codon was TAA, TGA, and TAG; and (iii) the ORF was more than 100 amino acids in length. TestCode analysis, in the GCG Wisconsin package, was used to identify those ORFs most likely to express a gene. ORFs likely to code for a protein were given the designations MS100 to MS163, starting with MS100 and proceeding clockwise as shown on the map in Fig. 1. Putative ORFs were compared to the protein databases by using BLASTX (1). Only searches that returned E values of <10⁻⁵ were considered sufficiently significant to be reported in Table 1.

FIG. 1.

Graphical map of plasmid pSD15. The putative ORFs identified on pSD15 are indicated by arrows. ORFs related to agarases are red, those related to mobile elements are yellow, ORFs similar to noncharacterized proteins are purple, ORFs with no homologs in GenBank are green, and all other ORFs are dark blue. The ORFs are more fully described in Table 1. Also indicated on the map are the large inverted repeats, IR-I and IR-II, putative DnaA boxes, and the five iterons and an AT-rich region that may play a role in replication.

TABLE 1.

Putative pSD15 ORFs and their functions

pSD15 ORF	Gene position (nucleotide)		Strand	Size of putative protein (aa)	Function, closest match	GenBank accession no. of closest match	pSD15 ORF alignment region (aa, range)	Homolog alignment region [aa range (total aa)]	% Identity
	Start	Stop
MS100	871	1911	+	347	Replication protein (RepA), Riemerella anatipestifer	AAD33095	115–298	95–263 (309)	25
MS101	2210	3049	+	280	Partition protein (SOJ homolog), Pyrococcus abyssi	B75038	2–248	4–251 (257)	31
MS102	3502	4929	+	478	Replicative DNA helicase (DnaB), Bacillus stearothermophilus	AAD20314	34–473	14–438 (454)	36
MS103	5139	6377	+	413	Transposase (similar to ISRm3) from plasmid pMT1, Yersinia pestis	AAC13227	49–402	46–398 (402)	46
MS104	6997	8046	+	350	2-Keto-3-deoxygluconate kinase (KdgK), Thermotoga maritima	AAD35161	5–350	2–339 (339)	47
MS105	8173	8838	+	222	2-Dehydro-3-deoxyphosphogluconate aldolase/4-hydroxy-2-oxoglutarate aldolase (KdpG), Thermotoga maritima	AAD35160	14–209	8–199 (205)	38
MS106	9720	12986	+	1089	Outer membrane protein, Porphyromonas gingivalis	CAA10226	48–1089	2–1017 (1017)	25
MS107	13001	14716	+	572	ORF, no homolog
MS108	14742	16355	+	538	ORF, no homolog
MS109	16469	18676	+	736	β-Agarase, Pseudoalteromonas atlantica	P13734	29–485	26–482 (505)	38
MS110	18732	19955	+	408	Hypothetical protein from Streptomyces coelicolor A3(2)	CAB61805	54–407	7–362 (370)	50
MS111	19967	21235	+	423	ORF, no homolog
MS112	21352	21711	+	120	Na+/glucose symporter, Vibrio parahaemolyticus	BAA11215	9–117	1–109 (530)	78
MS113	21813	22124	+	104	Na+/glucose symporter, Vibrio parahaemolyticus	BAA11215	1–83	147–229 (530)	60
MS114	22054	22953	+	300	Na+/glucose symporter, Vibrio parahaemolyticus	BAA11215	2–300	231–530 (530)	56
MS115	23245	27234	+	1330	β-Agarase precursor, Pseudoalteromonas atlantica	P13734	25–524	26–482 (505)	33
MS116	27486	29327	+	614	β-Agarase B precursor, Cytophaga drobachiensis	AF098955	37–377	59–351 (353)	38
MS117	29826	31211	+	462	Maturase, Pseudomonas alcaligenes	AAB68949	45–452	55–464 (490)	43
MS118	31698	32120	+	141	Transposase of ISW1, Wolbachia sp.	BAA73610	1–137	1–137 (144)	49
MS119	33875	32256	−	540	Fucosidase, Thermotoga maritima	AAD35394	53–476	5–392 (449)	31
MS120	35465	33879	−	529	Fucosidase, Thermotoga maritima	AAD35394	45–483	8–415 (449)	30
MS121	35878	37314	+	479	ORF, no homolog
MS122	38281	37469	−	271	Esterase, Caldicellulosiruptor saccharolyticus	P23553	43–266	34–264 (266)	27
MS123	39417	38284	−	378	Alcohol dehydrogenase, Streptomyces coelicolor A3(2)	CAB46803	13–377	8–365 (365)	31
MS124	39589	40437	+	283	Transcriptional activator (MtlR), Pseudomonas fluorescens	AAC34292	18–276	22–290 (301)	24
MS125	40616	41773	+	386	Cytochrome P-450, Bacillus subtilis	P27632	75–364	86–385 (405)	31
MS126	41809	42114	+	102	Ferredoxin, Caulobacter crescentus	P37098	1–100	1–101 (106)	45
MS127	42170	43420	+	417	Ferredoxin reductase, Pseudomonas putida	P16640	9–415	8–417 (422)	41
MS128	44151	43609	−	181	ORF, no homolog
MS129	45397	44336	−	354	Hypothetical protein, Mycobacterium tuberculosis	CAB06436	48–351	8–295 (299)	41
MS130	47228	45738	−	497	β-Agarase, Pseudoalteromonas atlantica	P13734	1–237	241–467 (505)	35
MS131	47791	47267	−	175	Hypothetical protein (YwoF), Bacillus subtilis	CAB05379	51–127	31–107 (468)	41
MS132	48975	48112	−	288	β-Agarase, Streptomyces coelicolor	P07883	31–286	44–309 (309)	24
MS133	49678	51213	+	512	Arylsulfatase A, Mus musculus	P50428	28–428	15–372 (506)	32
MS134	51265	51870	+	202	Arylsulfatase E, Homo sapiens	NP_000038	31–185	30–189 (589)	39
MS135	51955	52701	+	249	Arylsulfatase B, Mus musculus	ARSB_MOUSE_2	1–153	99–252 (252)	36
MS136	54108	52855	−	418	Hypothetical protein, Staphylococcus aureus	BAA86649	1–418	30–464 (507)	35
MS137	54073	54624	+	184	Dehydrogenase, Haemophilus influenzae	Q57517	15–183	169–342 (342)	26
MS138	54975	55412	+	146	Dehydrogenase, Synechocystis sp.	BAA18061	4–137	106–243 (249)	36
MS139	58035	56464	−	524	GGDEF family protein, Deinococcus radiodurans	F75583	37–154	74–180 (533)	24
MS140	58243	60804	+	854	Beta-galactosidase, Thermoanaerobacter ethanolicus	P77989	37–436	7–384 (743)	28
MS141	60926	61942	+	339	Dehydrogenase, Haemophilus influenzae	Q57517	29–328	33–332 (342)	28
MS142	61972	62739	+	256	Gluconate dehydrogenase, Bacillus subtilis	P50842	5–256	6–254 (254)	51
MS143	62770	64227	+	486	Aldehyde dehydrogenase, E. coli	P25553	3–483	1–478 (479)	42
MS144	64230	65357	+	376	Racemase, Pseudomonas putida	443131	113–361	110–355 (357)	33
MS145	65362	66414	+	351	Aldose epimerase, Acinetobacter calcoaceticus	P05149	6–343	26–377 (381)	38
MS146	66414	67415	+	334	l-Rhamnose transporter, E. coli	P27125	4–330	8–341 (344)	19
MS147	67545	69335	+	597	Chemotaxis protein, Bacillus subtilis	P39217	195–553	295–653 (662)	26
MS148	69402	69731	+	110	ORF, no homolog
MS149	70199	69765	−	145	ORF, no homolog
MS150	70934	70287	−	216	ORF, no homolog
MS151	71423	70947	−	159	ORF, rather weak similarity to HigB, proteic killer active protein from plasmid Rts1 (JC4694, 29% over 77 amino acids)
MS152	74351	75130	+	260	ORF, no homolog
MS153	75148	75906	+	253	ORF, no homolog
MS154	75920	76501	+	194	ORF, no homolog
MS155	77927	76518	−	470	Mobilization protein (MocB), Bacteroides fragilis	B48487	71–206	61–213 (318)	29
MS156	85243	86772	+	510	Transposase, Staphylococcus aureus prophage phiPV83	NP_061635	34–510	27–520 (548)	36
MS157	87182	87874	+	231	Hypothetical protein, Lactobacillus bacteriophage phi-adh	CAB52516	118–227	25–131 (170)	23
MS158	89336	90091	+	252	Hypothetical protein, Pseudomonas aeruginosa	AAG03933	1–235	1–232 (247)	29
MS159	91015	92565	+	517	Transposase, Pseudomonas alcaligenes	AAB48471	10–516	9–508 (509)	34
MS160	92534	93319	+	262	Transposase, Rhizobium sp. strain NGR234	P55617	25–258	6–239 (245)	41
MS161	94522	93842	−	227	HsdS of type I restriction-modification system, E. coli	CAA10700	1–197	8–205 (463)	22
MS162	95628	94525	−	368	Hypothetical protein, Xylella fastidiosa	AAF84517	15–342	10–336 (371)	26
MS163	100016	98931	−	362	HsdM of type I restriction-modification system, Xylella fastidiosa	AAF85527	6–362	16–351 (519)	44

Direct sequence repeats (iterons) were detected by using a modified version of the program Internal Repeat Finder, which was created and modified by Matteo Pellegrini (http://www.doe-mbi.ucla.edu/people/matteo/repeats.html). The G+C content was determined by using MacVector, while the GC profile was obtained with the BioPlot module of Vector NTI with a window size of 500. SignalP V1.1 (http://www.cbs.dtu.dk/services/SignalP/) was used to identify possible signal sequences in the putative agarases encoded by pSD15.

The GenBank nucleotide sequence accession number for pSD15 is AF339846.

RESULTS AND DISCUSSION

Morphology and growth characteristics of isolate PRE1.

Isolate PRE1 is a gram-negative rod-shaped bacterium ca. 6 μm in length. Colonies growing on M10 or M7 solid medium are pigmented orange and exhibit a strong agar-degrading activity. The strain grows optimally at 30°C, more slowly at 25°C, very poorly at 37 or 42°C, and not at all at 45°C. Cultures grown on solid medium at 30°C lose their viability very rapidly upon storage at room temperature.

Salt-dependency testing indicates that isolate PRE1 is truly a marine bacterium, requiring NaCl for growth. Growth occurs over a range of 0.5 to 6% NaCl, but significant growth only occurs at between 1.5 and 3% NaCl. K⁺ can be omitted from artificial seawater but Mg²⁺ and Ca²⁺ are indispensable for growth.

The 16S rRNA gene is a useful target for establishing prokaryotic phylogeny based on sequence similarity (25). Analysis of the sequence obtained for isolate PRE1 by using three conserved primers indicated that it is a Microscilla sp. in the Flexibacter group belonging to the Flexibacter-Cytophaga-Bacteroides phylum. The characteristics of PRE1, as described above, also fit the description of Microscilla species given in Bergey's Manual of Determinative Bacteriology (7).

Microscilla sp. strain PRE1 was resistant to gentamicin, kanamycin, neomycin, streptomycin, ampicillin, and trimethoprim and sensitive to rifampin, tetracycline, erythromycin, carbenicillin, chloramphenicol, nalidixic acid, and novobiocin at the concentrations tested. An identical pattern of resistance and sensitivity was observed for the strain cured of the pSD15 plasmid. This suggested the absence of any common antibiotic resistance determinants on pSD15, which was later confirmed by the sequencing results.

Agarase activity and plasmid stability.

The ability to degrade agar is a striking feature of Microscilla sp. strain PRE1 grown on solid medium. During the initial culturing of the organism, colonies with the same morphology but unable to degrade agar appeared at a relatively high frequency. Analysis of cells from these colonies by Gram staining, microscopic observation, antibiotic susceptibility, and 16S rDNA analysis indicated that they were identical to Microscilla sp. strain PRE1. The possibility that the loss of agar-degrading ability was related to loss of the native plasmid, pSD15, was therefore examined.

Several colonies which were no longer able to degrade agar were screened for plasmid content. While the native 101-kb plasmid was detected in the original parent, plasmid DNA was not detected in any of the strains which lost the ability to degrade agar (Fig. 2A; strains unable to degrade agar were designated PRE1-0). To confirm that plasmid pSD15 was indeed missing from these strains and did not integrate into the chromosome, we screened for its presence by PCR of total DNA by using a pair of primers specific to pSD15. To do so, a clone was selected from the library prepared for sequencing pSD15, and both ends of the ca. 2-kb insert were sequenced. On the basis of this sequence, two primers were designed and used to PCR amplify DNA from cell lysates prepared by boiling a suspension of cells from several colonies of Microscilla PRE1 or PRE1-0. As shown in Fig. 2B, a PCR fragment of the expected size was obtained from colonies able to degrade agar, while no product appeared from colonies unable to degrade agar. This confirmed that the nondegrading isolates arose from the parent Microscilla sp. strain PRE1 after loss of the native 101-kb plasmid, pSD15.

FIG. 2.

Plasmid pSD15 is not present in non-agar-degrading isolates. (A) Total DNA isolated from Microscilla colonies that had lost the ability to degrade agar (lane 1; PRE1-0) and from colonies capable of degrading agar (lane 2; PRE1). Lane 3 is DNA isolated from Rhizobium meliloti, harboring three plasmids, used as a molecular mass marker with sizes as indicated. The position of chromosomal DNA on the gel is also indicated (chr). (B) PCR amplification to identify the presence of a 1.5-kb pSD15 fragment in PRE1 colonies (lane 2), PRE1-0 colonies (lane 3), supercoiled pSD15 DNA (lane 4), a library clone carrying this fragment (from which primers were designed) (lane 5), and a library clone selected at random (lane 6). Lane 7 is a no-DNA template control. Lane 1 has a molecular mass marker (1-kb ladder; Gibco).

The stability of pSD15 in Microscilla sp. strain PRE1 was then tested by using the ability to degrade agar as a phenotypic marker for the presence of the plasmid. Colonies from mid-exponential-, late-exponential-, and stationary-phase cultures were checked for the ability to degrade agar on M10 medium as described in Materials and Methods. On average 2% of the cells at each stage of growth had lost the plasmid. This is a surprisingly high loss rate even for a low-copy-number plasmid. Most naturally occurring plasmids are lost at frequencies of 10⁻⁵ to 10⁻⁴ per generation. As discussed below, sequence analysis of pSD15 did not provide any evidence for a stabilization mechanism (such as partitioning, postsegregational killing, or conjugal transfer), which would help to limit its loss from the host. It is possible that this high loss rate is a result of culturing the strain in the lab as opposed to growth in the natural environment, where the ability to degrade agar (or any other traits conferred on the host by the plasmid) may provide enough of a selective pressure to maintain the plasmid. Another possibility is that Microscilla sp. strain PRE1 is not the “natural” host for pSD15. However, it should be noted that the putative replication initiation protein of pSD15, MS100, is very similar to replication initiation proteins of plasmids isolated from Riemerella anatipestifer, another member of the Flexibacter-Cytophaga-Bacteroides group of bacteria (see Table 1).

Sequence analysis.

pSD15, the single plasmid found in Microscilla sp. strain PRE1, is a circular plasmid of 101,648 bp (Fig. 1). It has a total of 98 ORFs larger than 100 amino acids, of which 64 are predicted to code for a protein as determined by TestCode analysis. Of these 64 putative proteins, 46 are coded for on one strand and 18 are coded on the other. Possible functions of these putative proteins, based on amino acid sequence similarities, are given in Table 1.

Analysis of the base composition of pSD15 revealed an overall G+C content of 42%. The chromosomal G+C content of Microscilla furvescens, Microscilla sp. strain PRE1's closest match in the RDP database, is reported to be 44% (7). As shown in Fig. 3, except for an AT-rich region located in the putative replication origin, the G+C content is distributed evenly throughout pSD15.

FIG. 3.

Percent G+C profile of pSD15. The GC profile was obtained by the BioPlot module of Vector NTI (version 5.5) with a window size of 500.

In general, pSD15 is a catabolic plasmid, conferring the ability to degrade and utilize polysaccharide polymers such as agar. A small part of the plasmid is dedicated to autonomous plasmid replication functions, and there are some additional putative functions, such as a DNA modification system and several transposases. In addition, ca. 25% of the plasmid, from about position 78000 to position 101648, has little coding sequence. Of the eight ORFs in this region (designated MS156 to MS163), three (MS156, MS159, and MS160) are similar to transposases, two appear to encode a restriction modification system (MS161 and MS163), and three are similar to proteins of unknown function from other organisms.

Analysis of putative agarase genes.

The most remarkable feature of pSD15 is the presence of five different genes potentially encoding agarases. Agar-degrading bacteria are ubiquitous in marine environments but are found as well in freshwater, sewage, and soil. These bacteria belong to diverse genera, including Alterococcus (38), Bacillus (23), Pseudomonas (17), and Microscilla (31), as well as those cited in Table 2. Bacteria capable of fixing nitrogen while growing on agar (39) or growing anaerobically with agar as the sole carbon source (38) have been isolated. Agarases have been purified and characterized from several different bacteria (2, 3, 23, 31, 36, 44, 46, 49), and in several cases the proteins have been overexpressed (24, 34, 35). Several of the agarase-encoding genes have also been cloned but no organism has been found to encode more then two agarases (Table 2).

TABLE 2.

Previously sequenced agarase genes

Agarase gene	Organism	Putative protein (no. of aa)a		Type of agarase	Accession no.
		Signal peptide	Mature peptide
aagA	Aeromonas sp.	–b	290	β-Agarase	AAF03246
agaA	Alteromonas agarilytica	26	1,403	α-Agarase	AAF26838
agaA	Cytophaga drobachiensis	19	520	β-Agarase	AAF21820
agaB	Cytophaga drobachiensis	18	335	β-Agarase	AAF21821
agrA	Pseudoalteromonas atlantica	23	482	β-Agarase	P13734
dagA	Pseudoalteromonas atlantica	21	269	β-Agarase	M73783
dagA	Streptomyces coelicolor	30	279	β-Agarase	CAA29257
agaA	Vibrio sp.	–	995	β-Agarase	BAA03541
agaB	Vibrio sp.	–	955	β-Agarase	BAA04744

^aaa, amino acids. 

^b–, no signal peptide reported. 

The agarase-encoding genes of pSD15, MS109, MS115, MS116, MS130, and MS132 are dispersed in a 32.5-kb region from bp 16469 to bp 48975. An alignment of the deduced amino acid sequence for the five genes was performed by using Vector NTI after the introduction of gaps to compensate for size differences between the proteins. The analysis did not identify any region of homology common to all five putative agarases. However, the C-terminal two-thirds of MS109 (from amino acids 241 to 736) showed 32.7% identity to MS130, while the N-terminal half of MS116 (amino acids 1 to 311) showed 18.4% identity to MS132. The method of Belas (6) was also used to compare the predicted protein sequences for the five genes. The amino acid sequences were first simplified by categorizing each residue according to the general functional properties: (i) neutral, weakly hydrophobic (P, A, G, S, or T residues); (ii) hydrophilic, acid amine (Q, N, E, and D); (iii) hydrophilic, basic (H, K, and R); (iv) hydrophobic (L, I, V, and M); (v) hydrophobic, aromatic (F, Y, and W); or (vi) cross-link forming (C). Again, there was no single region common to all five proteins.

The five putative agarases were then compared to the deduced amino acid sequence data for nine different agarase genes available in public databases (described in Table 2). As shown in Table 1, three putative agarases from pSD15, MS109, MS115, and MS130 match with the agrA-encoded protein from Pseudoalteromonas atlantica. MS116 is most similar to the β-agarase B precursor of Cytophaga drobachiensis, while MS132 is most similar to β-agarase from Streptomyces coelicolor. When the deduced amino acid sequences for all 14 agarase proteins were aligned (five from pSD15 plus the nine previously reported), only a few randomly distributed residues were found to be conserved in all 14 proteins (data not shown). This result agrees with a prior conclusion that microorganisms appear to degrade agar by using a series of enzymes with narrow specificities rather than a single enzyme with a broad specificity (43).

A signal peptide is common to most secreted proteins. Six of the nine previously reported agarases contain a signal peptide of 18 to 30 amino acids in length (Table 2). Analysis of the five putative agarases encoded by pSD15 by using the SignalP program resulted in the identification of a signal peptide within the first 40 amino acids for four of the proteins (Table 3) (32).

TABLE 3.

Signal peptides identified in the N-terminal 40 amino acids of putative agarases encoded by pSD15^a

ORF	Sequence
MS109	MRIKIWISYGLCFILMIFLGSDILA↓QKVEVDVQFNVKHVV
MS115	MYGKVVLFTVLFLGNIFCLYSQ↓GVQ↓VDVNLNVKHSVGGVS
MS116	MKSPIVIYYFTKKPSNMKTHLTLLITWIAFLGAKA↓QDWSG
MS132	MKKTYLGLALLFIAYQSTLAQS↓QPTVNEGEPVAQLEWELV

^aNo signal peptide was identified in MS130. The downward arrow indicates the position(s) of the predicted cleavage site(s). 

Previous studies have indicated that there are two pathways for the degradation of agar, which is a linear sulfated galactan composed of two regularly repeated galactose units alternatively linked by β-d-(1,4) and α-l-(1,3) linkages. The β-agarases cleave the β-d-(1,4) linkages, while the α-agarases cleave α-l-(1,3) linkages. As shown in Table 2, most purified agarases have been identified as β-agarases, with only the Alteromonas agarilytica enzyme being an α-agarase. The putative agarases encoded by pSD15 on the basis of amino acid sequence are most similar to β-agarases; however, confirmation of this must await purification and biochemical analysis of the activity of these proteins.

Replication and plasmid maintenance.

The region from position 1 to bp 4929 of pSD15 has an organization typical of many prokaryotic replicons (19). It includes five direct repeats (putative iterons), an ORF (MS100) encoding a putative 374-amino-acid replication initiation protein, and a 316-bp AT-rich region (86% AT; positions 1916 to 2231). The five iterons share a 16-bp consensus sequence and are separated by 73, 65, 69, and 7 bp, respectively (Fig. 4). Immediately downstream of the AT-rich region are the coding regions for two proteins, MS101 and MS102, which share similarities, respectively, to a component of the partitioning system from Borrelia burgdorferi and the replicative helicase, DnaB, of Bacillus stearothermophilus. Two putative DnaA box sequences, TTATCCACA and TGTGGATAA, are distantly separated from each other, and both boxes were located a considerable distance from the origin of replication, making them unlikely to be involved in replication (see Fig. 1).

FIG. 4.

Putative replication region of pSD15. (A) A 5-kb region of pSD15 contains a putative replicase (MS100), helicase (MS102), partition protein (MS101), five iterons, and a 316-bp AT-rich region. (B) The sequence of the five iterons is shown, with the base-pair position on pSD15 indicated in parentheses. Mismatched bases are indicated in boldface.

In an attempt to clone a fragment of pSD15 capable of autonomous replication, a ca. 5-kb fragment containing the five iterons, the AT-rich region, and ORFs MS100 to MS102 was obtained by PCR and cloned into pBR325 (which requires PolA for its replication). Attempts to introduce this construct, designated pZZ3, into a polA E. coli strain by electroporation failed, suggesting that the cloned replicon could not function in E. coli. Since pBR325 can be mobilized (4), attempts were then made to conjugally transfer pZZ3 into PRE1-0 (the Microscilla isolate cured of the native plasmid) by using E. coli SM10 (40) (which has the Tra⁺ plasmid RP4 integrated in the chromosome) carrying pZZ3 as the donor. We also failed to obtain any exconjugates, which suggests either that the cloned fragment doesn't contain all regions required for autonomous replication or that the selective marker on pZZ3, chloramphenicol, is not expressed and/or functional in Microscilla sp. strain PRE1.

No evidence for tra-type genes, which would encode the elements necessary for conjugal transfer, was found on pSD15. Furthermore, attempts to transfer pSD15 from the parent to a spontaneous rifampin-resistant derivative of the Microscilla PRE1-0 strain failed. The only ORF encoded by pSD15 that has homology to a protein involved in plasmid transfer is MS155. This protein has significant homology with MocB (for mobilization cassette) of Tn4399 isolated from Bacteroides fragilis. Derivatives of nonconjugal plasmids that carry Tn4399 can be mobilized in trans by the broad-host-range IncP plasmid pRK231 or R751 in E. coli (29). This mobilization requires a 2.8-kb region of Tn4399 that encodes both the MocA and MocB proteins. A homolog of MocA was not found on pSD15.

MS161 and MS163 are similar to HsdS of E. coli and HsdM of Xylella fastidiosa, respectively. These are two components of the type I family of restriction-modification systems (see reference 30 for a review). HsdS is the specificity subunit, and HsdM is the modification subunit. An often-found third component, HsdR, which is the restriction subunit, was not identified on pSD15. However, hsdS and hsdM usually form an operon unlinked to hsdR. While HsdR is necessary for restriction, HsdM and HsdS by themselves are sufficient for modification methylation. Type II restriction-modification systems have been shown to function as a postsegregational killing system for plasmid stabilization (18), but there is no report of the type I system having similar properties.

Polysaccharide utilization.

A number of genes dedicated to polysaccharide or oligosaccharide catabolism were found on pSD15. MS106 is most similar to an outer membrane protein from Porphyromonas gingivalis, and it shows significant identity to an outer membrane protein from Bacteroides thetaiotaomicron (accession number JC6027) that is essential for the utilization of malto-oligosaccharides and starch (data not shown). An attractive hypothesis is that MS106 enables Microscilla sp. strain PRE1 to import the oligosaccharides that are generated by the activities of its extracellular agarases.

MS104 is similar to 2-keto-3-deoxygluconate kinase, which is a component of the glucuronic acid catabolism pathway. Glucuronic acid is one of the building blocks of glycosaminoglycans, a very abundant heteropolysaccharide. Glucuronate catabolism genes are usually part of catabolic pathways of polysaccharide degradation. MS105 putatively encodes 2-dehydro-3-deoxyphosphogluconate aldolase (KdpG). This is a key enzyme in the Entner-Doudoroff pathway and participates in the regulation of the intracellular level of glyoxylate. MS119 and MS120 are predicted to encode proteins with sequence similarity to a fucosidase. These two proteins themselves are 64.4% identical. MS122 is similar to an esterase that is involved in xylan degradation. MS140 is predicted to be a β-galactosidase. MS145 is similar to aldose epimerase, a mutarotase active in galactose metabolism. In addition, six putative dehydrogenase genes were found. MS123, MS137, MS138, and MS141 appear to be alcohol dehydrogenases, MS142 putatively is a gluconate dehydrogenase, and MS143 putatively is an aldehyde dehydrogenase.

Transposases and insertion sequences.

Five ORFs with homology to transposases—MS103, MS118, MS156, MS159, and MS160—were found in pSD15. Inverted repeats (with mismatches underlined) could be identified near each of these ORFs. MS159 and MS160 are flanked by inverted repeats, which start at positions 90875 (5′-GTGCCAGTGATTACGG) and 93368 (5′-CCGAAATCACTGGCAC). MS156 is flanked by 5′-TCTTTTTTT (starting at position 85165) and 5′-AAAAAAAGA (starting at position 88860). The inverted repeats of MS118, 5′-TTTGCGATATTTC, start at position 31682 (just upstream of the ATG start of MS118) and at position 31775 (within the coding sequence). There were two sets of inverted repeats flanking MS103, each located within another ORF (positions 3813 [5′-TGGAGCTGC] and 9813 [5′-GCAGCTCCA] and positions 3490 [5′-CTTTCTTCTTTTCCTGAA] and 10204 [5′-TTCAGAAAAAGAAGGAAG]).

There are two additional large inverted repeats in pSD15. These repeats, termed IR-I and IR-II, are 99.6% identical. IR-I is 1,381 bp long, starts at position 54124, and ends at position 52744. IR-II starts at position 85503, ends at position 86886, and is 1,384 bp long. These inverted repeats overlap MS136 and MS156.

Additional possible functions.

There are three ORFs—MS112, MS113, and MS114—that show similarities to different regions of the 531-amino-acid Na⁺/glucose symporter encoded by the sglS gene of Vibrio parahaemolyticus, a slightly halophilic marine bacterium (37). MS112 (120 amino acids) matches amino acids 1 to 109 of the V. parahaemolyticus symporter with 78% identity. MS113 (104 amino acids) matches residues 147 to 229 with 60% identity and MS114 (300 amino acids) matches residues 231 to 530 with 56% identity. It is possible that pSD15 encodes the symporter in three parts, which are assembled together to form a functional unit.

Na⁺/glucose transporters have been shown to utilize a Na⁺ electrochemical gradient to drive sugar transport in mammalian cells (33), but except for the SglS transporter from V. parahaemolyticus, this is unusual in bacteria (37). A conserved amino acid sequence (G… AXXXXLXXXGR) has been proposed to be involved in Na⁺ recognition or binding and has been found in several Na⁺-coupled symporter proteins (11, 48). This motif was found in MS114 from residues 110 to 158 (G₁₁₀… A₁₄₈XXXXLXXXGR₁₅₈) with the motif of SglS from V. parahaemolyticus (G₃₃₉… S₃₇₇XXXXLXXXGR₃₈₇) being very similar (37). Since the first 83 residues of MS113 show 60% identity (78% similarity) with residues 147 to 229 of SglS, which have been shown to be involved in sugar recognition or transport (37), it is likely that MS113 contributes to sugar recognition or transport.

Upstream of the putative transposase MS118, an ORF (MS117) showed 43% identity to a maturase-related protein of a group II intron from Pseudomonas alcaligenes (50). Group II introns found in bacteria are almost always associated with or inserted in another mobile genetic element (see references 13 and 27 for reviews). Maturase is an enzyme that promotes RNA splicing, which is commonly found in various introns, particularly in group II introns (12, 20, 28). Previous studies indicate that all known maturases are expressed from intronic ORFs fused to the upstream exon (20). Our sequence analysis, however, did not identify the presence of any potential exon(s) in proximity to this maturase-encoding ORF.

Concluding remarks.

Analysis of plasmid pSD15 has revealed a rich array of genes whose putative products appear to be involved in polysaccharide degradation and sugar transport and modification. This plasmid, while dispensable for growth of the bacterium under laboratory culturing, undoubtedly plays an important role in the adaptation of its host organism to its marine environment. This is reflected in the presence of five genes related to known agarases. While attempts to demonstrate conjugal transfer of the plasmid under laboratory conditions failed, the possibility of a conjugal transfer system unrelated to known systems cannot be ruled out, particularly in view of the surprisingly high frequency of spontaneous loss of the plasmid during culturing in the lab. Much remains to be done to understand the many novel features contributed by this plasmid to its host.