Chitin Catabolism in the Marine Bacterium Vibrio furnissii

Chitin catabolism by the marine bacteriumVibrio furnissii involves many genes and proteins, including two unique periplasmic hydrolases, a chitodextrinase and a β-N-acetylglucosaminidase (Keyhani, N. O., and Roseman, S. (1996) J. Biol. Chem. 271, 33414–33424 and 33425–33432). A specific chitoporin in the outer membrane may be required for these glycosidases to be accessible to extracellular chitooligosaccharides, (GlcNAc) n , that are produced by chitinases. We report here the identification and molecular cloning of such a porin. An outer membrane protein, OMP (apparent molecular mass 40 kDa) was expressed when V. furnissii was induced by (GlcNAc) n , n = 2–6, but not by GlcNAc or other sugars. Based on the N-terminal sequence of OMP, oligonucleotides were synthesized and used to clone the gene,chiP. The deduced amino acid sequence of ChiP is similar to several bacterial porins; OMP is a processed form of ChiP. InEscherichia coli, two recombinant proteins were observed, corresponding to processed and unprocessed forms of ChiP. A null mutant of chiP was constructed in V. furnissii. In contrast to the parental strain, the mutant did not grow on (GlcNAc)3 and transported a nonmetabolizable analogue of (GlcNAc)2 at a reduced rate. These results imply that ChiP is a specific chitoporin.

We have previously reported that the chitin catabolic cascade and signal transduction systems expressed by the marine bacterium Vibrio furnissii comprise a large number of genes and proteins, only some of which have been identified (6 -15). 1 Two of these proteins are unique periplasmic glycosidases. One is a chitodextrinase (11), and the other is a ␤-N-acetylglucosaminidase (12). The concerted action of these two enzymes yields GlcNAc and (GlcNAc) 2 from the higher oligosaccharides, which are then taken up by specific cytoplasmic membrane transporters (9,10,13).
But how do the higher oligosaccharides penetrate the outer membrane/cell wall complex so that they can be hydrolyzed in the periplasm? There is an extensive literature on outer membrane proteins or porins that are thought to mediate this process (16 -18). The nonspecific, constitutive porins of Escherichia coli permit the diffusion of solutes that range up to about 600 Da (19), depending on shape, or roughly the size of a trisaccharide. A few sugar-specific porins, or glycoporins, have been reported. These include, most notably, the LamB protein or phage receptor protein, whose crystal structure has been resolved (20) and which permits the diffusion of malto-oligosaccharides (21,22), and ScrY, involved in sucrose transport (23). The RafY protein is involved in raffinose uptake (24) and may not have a specific binding site for the trisaccharide but functions because it is wider than the constitutive E. coli porins (25).
The present report presents evidence for another glycoporin designated chitoporin. The porin is expressed by V. furnissii in its outer membrane and is induced by chitin oligosaccharides but not by GlcNAc nor by other sugars. The structural gene for the porin, chiP, has been cloned, and its sequence was determined and expressed in E. coli. The physiological behavior of a null mutant of chiP in V. furnissii provided additional evidence that ChiP is a specific chitoporin.

Growth and Maintenance of Strains
The strains and plasmids used are given in Table I. E. coli strains were grown in LB or on Luria Agar plates supplemented with 50 -75 g/ml ampicillin where appropriate for selection of recombinants. Cell * This work was supported by Grant GM51215 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM 1 The subject matter of the accompanying manuscripts is as follows: (GlcNAc) 2 , a PTS sugar in E. coli (1); characterization of IIA Chb from E. coli (2); characterization of phospho-IIB Chb and of a potential transition state analogue in the phosphotransfer reaction between IIA Chb and IIB Chb of E. coli (3); analytical sedimentation studies on IIA Chb , IIB Chb , the phosphoproteins, and a model transition state analogue (4); and cloning and characterization of a (GlcNAc) 2 phosphorylase from V. furnissii (5).
cultures were grown at 37°C, with aeration and turbidity measured at 600 nm. At this wavelength, 1.0 optical density unit corresponds to 0.5 mg of cell protein/ml. V. furnissii strains were grown either in high salt LB (LMB, Luria broth supplemented with an additional 10 g/liter NaCl) or in minimal media containing HEPES (50 mM, pH 7.5), 50% artificial sea water (ASW), 0.1% NH 4 Cl, 0.001% K 2 HPO 4 , and 0.5% DL-lactate (lactate-ASW) (8). The minimal medium was supplemented with other carbon sources as indicated. Cell cultures were grown at 30°C with aeration and growth measured by absorbance at 540 nm; 1.0 optical density corresponds to 0.5 mg of cell protein/ml. Growth curves were obtained by growing primary inocula overnight in LMB and diluting 1:100 in the indicated media.

DNA Manipulations
DNA preparation and analysis, restriction enzyme digests, ligation, and transformations were performed using standard techniques (28). A cosmid library was constructed using bacterial genomic DNA from V. furnissii 1514 as described (28). Library construction, including conditions for partial genomic DNA restriction (using Sau3AI) and ligation into the cosmid vector Supercos1, was performed as recommended (Stratagene). The ligation mixture was packaged into phage using GigaPack Gold III packaging extract (Stratagene). Transfections into various E. coli strains were performed according to the supplier's recommendations. Double-stranded DNA was prepared from recombinant clones and sequenced by the dideoxy method using U.S. Biochemical Sequenase version 2.0 sequencing kit or alternately by the Genetics Core Facility (Johns Hopkins Medical School) using an ABI-373 automated sequencer.

Isolation of Outer Membranes
Unless otherwise indicated, the following procedures were conducted between 0 and 4°C.
Method 1-Outer membrane fractions were prepared from mid-exponential phase cells (29,30). Typically, 5-10-ml cultures were harvested (7500 ϫ g, 5 min), washed once with the same volume of buffer (50 mM Tris-HCl buffer, pH 7.5, 50 mM EDTA, 15% sucrose), and then resuspended in 0.5 ml of buffer containing 0.3 mg/ml lysozyme. Samples were maintained on ice for 30 min in a microcentrifuge tube and centrifuged at maximum speed for 15 min. The resulting pellet was resuspended in 1 ml of ice-cold 1% N-lauryl sarcosine to solubilize the inner membrane fraction. The sample was passed through a 22-gauge needle 5-10 times and centrifuged for 15 min, and the supernatant fluid was discarded. The pellet was washed either with 1% lauryl sarcosine or with H 2 O. SDS-PAGE loading buffer (20 -100 l) was added to the pellet, and samples were boiled for 5 min before being subjected to SDS-PAGE (28).
Method 2-Outer membrane enriched fractions were prepared from mid-exponential phase cells that were first treated by sonication in 50 mM Tris-HCl buffer, pH 7.5. N-Lauryl sarcosine was then added to a final concentration of 0.5%, and the mixture was incubated at 25°C for 30 min (31). Samples were centrifuged at 6000 ϫ g for 2 min; the supernatant fraction was then centrifuged in a Beckman Airfuge at 150,000 ϫ g for 30 min. The pellet was resuspended in SDS-PAGE loading buffer and processed as described above for Method 1.
Method 3-Inner and outer membranes from V. furnissii were pre-pared essentially as described (32). Briefly, mid-exponential cells, grown in lactate-ASW medium with and without 0.6 mM (GlcNAc) 2 , were harvested at 12,000 ϫ g, 10 min, washed with 50 mM Tris-HCl buffer, pH 7.5, containing 50 mM EDTA and 15% sucrose, and resuspended in the same buffer containing 0.3 mg/ml lysozyme. Samples were maintained on ice for 30 min, then slowly poured into four volumes of ice-cold sterile distilled H 2 O with rapid stirring, and stirred for an additional 10 min. Unlysed cells were removed by centrifugation at 1200 ϫ g for 15 min. The supernatant was centrifuged at 360,000 ϫ g for 2 h, and the pellet was resuspended in the same volume of buffer without lysozyme and passed through a 22-gauge needle 5-7 times. The membranes were harvested at 360,000 ϫ g, resuspended in 25% sucrose containing 5 mM EDTA, pH 7.5, layered on top of a 30 -55% sucrose gradient, and centrifuged at 180,000 ϫ g for 16 h. Fractions were collected by piercing the bottom of the centrifuge tubes and allowing the liquid to flow out. The buoyant density of each fraction was determined from measurements of the refractive index.

Lipopolysaccharide Estimation (KDO Assay)
Gradient fractions from the inner/outer-membrane isolation procedure were assayed for 3-deoxyoctulosonic acid (KDO) as described (32). Gradient fractions (100 l) were precipitated with 1 ml of ice-cold 10% trichloroacetic acid. The pellet was washed twice with 1 ml of distilled H 2 O, resuspended in 0.1 ml of 0.02 N H 2 SO 4 , and hydroyzed for 20 min at 100°C. The hydrolysates were analyzed for KDO using the thiobarbituric acid method.

Enzyme II Nag Assay
Fractions from bacterial inner and outer membranes were assayed for phosphoenolpyruvate-dependent GlcNAc phosphorylation catalyzed by the PTS system (33,34). Each assay mixture contained the following components in 0.2 ml: 0.1 ml of the membrane fraction to be assayed, 2-5 units of homogeneous Enzyme I, 3-5 M HPr, 125 g of bovine serum albumin, 50 mM Tris-HCl buffer, pH 8.0, 10 mM potassiumphosphoenolpyruvate, 5 mM MgCl 2 , 1 mM dithiothreitol, 10 mM KF, and 2 mM [ 14 C]GlcNAc (315 dpm/nmol). The assay mixtures were incubated for 15 or 30 min at 37°C and heated for 5 min at 100°C. GlcNAc-6-P was determined by anion exchange chromatography (35).

N-terminal Sequence of ChiP
The protein band observed on SDS gels was electroblotted to an Immobilon-P polyvinylidene difluoride membrane. After transfer, the blot was lightly stained with Coomassie Blue, and the band was cut out. The N-terminal sequence of the protein was determined at the Biosynthesis and Sequencing Facility (Department of Biological Chemistry,

Cloning of chiP
The putative chitoporin gene, chiP, was cloned using a two-step strategy. First, primers were designed based on the N-terminal amino acid sequence and used to clone the 90 base pairs corresponding to the nucleotide sequence of the N terminus. Then the N-terminal nucleotide fragment was used as a probe for screening a recombinant V. furnissii cosmid library in E. coli.
Three degenerate oligonucleotide probes were synthesized based on the N-terminal amino acid sequence of ChiP: (a) primer O1, GGCG-GAATTCAARGARGTNGGNGT; the sequence in bold is derived from amino acids 1-5, and an EcoRI site inserted for use in cloning is underlined; (b) Primer O2, GHGTSTAYGGNGTSGCSGCSATG (amino acids [12][13][14][15][16][17][18][19]; and (c) primer O3, TTRTGNCTRTTRCCTAGGGGG; the sequence in bold is derived from amino acids 25-29, i.e. direction of primer is opposite to O1, and a BamHI cloning site is underlined. Primers O1 and O3 were used to amplify a 90-base pair N-terminal fragment of chiP by polymerase chain reaction from V. furnissii genomic DNA. The polymerase chain reaction generated fragments were subcloned into the EcoRI-BamHI sites of vector pBluescript II KS(ϩ), and colonies were screened for those containing the correct insert by hybridization to primer O2. Three such isolates designated pKS-X1, pKS-X2, and pKS-X3 were picked, and their inserts were sequenced. All three contained the DNA sequence that encoded the desired amino acid sequence. The N-terminal nucleotide fragment (designated N90f) was cut from pKS-X1 (using EcoRI and BamHI) and random primer-labeled with [␣-32 P]dATP according to the manufacturer's recommendations (Amersham Pharmacia Biotech). The radiolabeled probe was used to screen a recombinant V. furnissii cosmid library and two positive clones, designated as pL7 and pL8, containing 25-30-kb inserts were isolated by colony hybridization from approximately 3000 recombinant clones. Restriction analysis showed that both isolates shared many similar restriction fragments and a 4.2-kb EcoRI-HindIII fragment hybridized to N90f in both isolates as detected using a Southern blot. Attempts to subclone this fragment yielded a variety of deletions, and the intact 4.2-kb piece could not be cloned. The putative porin gene was, therefore, sequenced by subcloning fragments of the 4.2-kb piece using a combination of various restriction enzymes and polymerase chain reaction. Based on the sequences of the fragments, it was possible to derive the nucleotide sequence of the complete gene. This deduced sequence was confirmed by designing the appropriate primers and resequencing the entire 4.2-kb fragment using the original template (pL7). The chiP open reading frame was subsequently subcloned into the expression vectors pET21a and pIH1148 using appropriate primers based on the nucleotide sequence.

Construction of chiP Deletion Mutant
A knock-out or null mutant of chiP was constructed in V. furnissii by homologous recombination between a cloned fragment of chiP (constructed in the suicide vector pNQ705) and the V. furnissii genome (36). Briefly, this method involves conjugal transfer of plasmids from an E. coli mobilizing donor (strain S17-1) to V. furnissii. A fragment of the target gene is subcloned into the vector pNQ705, which contains an antibiotic resistance marker and is capable of propagating only in the E. coli host and not in the Vibrio (37). Conjugal transfer is achieved simply by mixing the two cell types together and then selecting against the E. coli host (amp s , V. furnissii is resistant up to 30 g/ml ampicillin) and for the selective marker (chloramphenicol resistance) on the (transferred) vector. Because the vector is incapable of propagating in V. furnissii, the chloramphenicol selective pressure results in recombination, giving the desired null mutant (36,38).
For the efficient construction of transformants with plasmids generated in E. coli, it was necessary to overcome the restriction barrier in V. furnissii. The restriction system was found to be similar to Sau96I (commercially available), and resistance to this system is effected in V. furnissii by a methylation modification system. To appropriately modify the plasmids, the V. furnissii methylase gene was cloned into E. coli strain S17-1, and plasmids were propagated in this strain prior to their use for transforming V. furnissii. 4

Suicide Vector Construction
A 0.3-kb fragment (NdeI-ScaI) of the N-terminal portion of chiP was isolated from pET-chiP, blunt ended using the appropriate nucleotides and Klenow fragment (New England Biolabs, Beverly, MA), and subcloned into the EcoRV site of pNQ705, yielding pNQ-X300.

Transconjugation of pNQ-chiP from E. coli to V. furnissii
The suicide plasmid, pNQ-X300, containing 0.3 kb of the porin gene was transferred into V. furnissii by transconjugation, using the conjugative E. coli strain S17-1. The latter was cotransformed with the suicide vector pNQ-X300 (which can propagate in this strain) and a compatible plasmid bearing the cloned V. furnissii DNA-methylase gene. The resultant double transformant was used in conjugation experiments with V. furnissii. Briefly, E. coli and V. furnissii, were inoculated and grown separately under appropriate conditions. V. furnissii 1514 was grown in 50 ml of LMB at 30°C to an A 540 ϭ 0.8; E. coli S17-1 cells harboring pNQ-X300 and pVfu129 (the methylase gene) were grown in 50 ml of LB at 37°C supplemented with 30 g/ml chloramphenicol and 50 g/ml kanamycin to an A 600 ϭ 0.8. Aliquots containing 2 ϫ 10 9 cells of each species were pipetted onto LMB plates (no antibiotics) and allowed to grow for 24 h at room temperature. The cocultured E. coli and V. furnissii cells were resuspended in 1.5-2.0 ml of LMB, and aliquots (0.1 ml) were transferred to LMB plates containing 30 g/ml ampicillin and 30 g/ml chloramphenicol. The desired V. furnissii recombinants are ampicillin-and chloramphenicol-resistant, individual colonies were picked, and a single colony was purified.

Complementation of V. furnissii chiP Null Mutant
The V. furnissii chiP gene was cloned into the mobilizable vector pSF4 (kind gift of Dr. V. N. Iyer, Carleton University, Ottawa, Canada) in a two step procedure. Primers were first constructed to clone chiP into the NdeI and BamHI sites of pIH1148 immediately following the Ptac promoter present on the plasmid. The fragment containing chiP fused to the Ptac promoter was then cloned into the PstI site of pSF4 using polymerase chain reaction and two newly designed primers. The primers used for the cloning were as follows: (a) for cloning into pIH1148: CW, 5Ј-CGGGCGCGCCATATGGACAAAATGTTTA-3Ј, and CCW, 5Ј-CGGGGATCCATTAGAAACCGTACTCTAGAC-3Ј and (b) for cloning into pSF4: CW, 5Ј-GGCCGAATGCATGTTTGACAGCTTATC-3Ј, and CCW, 5Ј-TAGTGATGCATAAGCTTGCCTGCAGGTCG-3Ј. The sequences in bold refer to NdeI, BamHI, NsiI, and NsiI sites on the given primers, respectively. It should be noted that NsiI and PstI result in compatible ends allowing for ligation into the PstI site of pSF4 giving pSF-chiP. pSF-chiP was tranferred into V. furnissii X1401 (chiP Ϫ null mutant) by transconjugation as described above.

Transport Assay
The rate of [ 3 H]Me-TCB uptake by V. furnissii was measured essentially as described (13,35,39). Briefly, V. furnissii was grown overnight in LMB and diluted 50-fold in lactate-50% ASW salts minimal media supplemented with 0.6 mM (GlcNAc) 2 (induced) or as indicated. Cells were grown at 30°C with aeration to an A 540 ϭ 0.8 -1.2, washed three times at 4°C with an equal volume of buffered 50% ASW salts, and resuspended in buffered 50% ASW salts or in 0.4 M sucrose containing 50 mM KCl, using 1 ⁄25 to 1 ⁄50 the volume of the growth medium. The suspension was stored on ice and transferred to room temperature 15 min prior to use. Transport experiments were conducted no later than 2 h after harvesting and washing of the cells. Uptake was initiated by the addition of an equal volume of cell suspension to [ 3 H]Me-TCB dissolved in the same buffer as the cell suspension. Substrate concentrations ranged from 0.5 to 100 M. The cell suspension was rapidly mixed at room temperature, and aliquots (0.1 ml) were taken at various times, added to 10 ml of wash (HEPES-buffered 50% ASW) at room temperature and filtered through Whatman GF/F glass microfiber filters. After washing with an additional 10 ml of buffer, the cells on the filter were solubilized with Packard Soluene-350 and counted in a Packard Liquid Scintillation Spectrometer.

Induction of an Outer Membrane Protein by Chitin Oligosac-
charides-Three different procedures were employed for the isolation of outer membrane enriched fractions from V. furnissii cells grown in lactate with 50% ASW with and without 0.6 mM (GlcNAc) 2. Fig. 1 illustrates the protein pattern from membranes prepared by Method 1. After induction by growth on (GlcNAc) 2 , a major new band was observed with an apparent molecular mass of 40 kDa, and representing 10 -20% of the total Coomassie-stained protein. Chitin oligosaccharides, (Glc-NAc) n , n ϭ 3-6, also acted as inducers, whereas the following sugars did not: GlcNAc, glucose, glycerol, maltose, melibiose, sucrose, trehalose, cellobiose, and glucosamine oligosaccharides, (GlcNH 2 ) n , n ϭ 1-3. Although GlcNAc did not act as an inducer when tested as described above, growth on GlcNAc as the sole carbon source (0.5%) did induce a band corresponding to ChiP, although at a greatly reduced level (Ͻ10% of that observed using 0.6 mM (GlcNAc) 2 as inducer). It should be noted, however, that most porins migrate at about the same molecular mass on SDS-PAGE, 30 -45 kDa (16 -18), so that the bands observed after growth on GlcNAc and after induction by (GlcNAc) 2 may or may not be the same proteins. Membrane proteins prepared by an alternate method (Method 2 under "Experimental Procedures") yielded essentially the same results as described above (data not shown).
To determine the location of the induced protein, and to establish that Methods 1 and 2 did indeed yield outer membranes, membranes were prepared from cells grown in lactate medium with and without 0.6 mM (GlcNAc) n , n ϭ 2-6, as inducer and lysed by osmotic shock, and inner and outer membranes were isolated by fractionation on sucrose gradients (Method 3). Gradient fractions were analyzed for KDO (outer membranes marker) and for the GlcNAc permease II Nag , inner membrane marker).
The results obtained with (GlcNAc) 2 -induced cells are shown in Fig. 2. Two KDO peaks are visible: a major peak at a sucrose density of about 1.220 -1.190 and a smaller peak at a density of 1.180 -1.160, as well as two peaks of Enzyme II Nag activity: a small amount of activity at a density of 1.180 -1.160 (corresponding to the smaller KDO peak) and another peak at a density of 1.140 -1.110. Aliquots of each fraction were analyzed by SDS-PAGE, and a 40-kDa band was observed in fractions corresponding to a density of 1.220 -1.190, with a small quantity detectable in fractions between densities of 1.180 and 1.160. All membrane preparations gave essentially the same profiles after isopyncnic sedimentation, but the 40-kDa protein was detected only in the KDO enriched fractions of induced cultures, i.e. cells grown in lactate ϩ 0.6 mM (GlcN-Ac) n , n ϭ 2-6. From these results, we concluded that the (GlcNAc) 2 -induced protein is localized in the outer membrane of V. furnissii.
Expression of ChiP by V. furnissii-Cultures were grown to mid-exponential phase in the presence of different concentrations of (GlcNAc) 2 , and the quantity of chitoporin or ChiP produced was measured as described under "Experimental Procedures." The optimal inducer concentration was 0.5-1.0 mM (GlcNAc) 2 (data not shown). Experiments on the time course of induction in the presence of 0.6 mM (GlcNAc) 2 showed that protein expression can be detected after 15 min of exposure to the inducer, with maximal expression occurring at approximately 2.5-3 h (Fig. 3). Cells washed into minimal media, without any carbon source after 3 h of induction, retained ChiP in the outer membrane for as long as 24 h after removal of the inducer.
N-terminal Determination of ChiP-The N-terminal amino acid sequence of a major V. furnissii outer membrane protein specifically induced by chitin oligosaccharides was sequenced by electroblotting the protein onto polyvinylidene difluoride membranes as described under "Experimental Procedures." The 32-amino acid residues determined by sequencing are given in Table II. The sequencing data confirmed that the band observed under (GlcNAc) 2 -induced conditions was a single polypeptide, Ͼ95% pure.
Cloning and Sequencing of chiP-The N-terminal amino acid sequence was used to construct a series of (degenerate) primers. These were used in turn to clone the gene from a V. furnissii library constructed in the cosmid vector SuperCos1, in a stepwise manner as described under "Experimental Procedures." Two cosmid clones were thus isolated with restriction analysis and hybridization, showing that they contained similar V. furnissii genomic inserts.
Initial attempts at subcloning the porin gene from the cosmid clone proved difficult, and although not fully confirmed, it appeared as if the gene immediately following the porin structural gene encodes a toxic protein. (When this segment of the DNA was sublconed from the cosmid into E. coli and expressed, the cells lysed. We presume that the cosmid was stably maintained because of low copy number and level of expression.) The sequencing and isolation of the gene encoding the porin was ultimately accomplished but required the "pasting together" of subfragments of the gene. The sequence derived from these fragments was confirmed by resequencing the gene using the original cosmid clone as the template and is presented in Fig. 4. The deduced amino acid sequence reveals that there are two potential start sites (methionine residues) separated by two amino acid residues. ChiP Expression in E. coli-The primary sequence of the chitoporin gene determined as described above was used to subclone the chitoporin gene into the overexpression vector, pET21a under the control of the T7 promoter ("Experimental Procedures"). Outer membrane proteins were analyzed by SDS-PAGE using Method 1 (Fig. 5). In Fig. 5, the position of migration of V. furnissii ChiP is marked by arrow 2; two recombinant proteins are expressed in E. coli, marked by arrows 1 and 2. The more rapidly migrating band (arrow 2), displays the same mobility as the V. furnissii ChiP. The two bands expressed in E. coli were electroblotted onto polyvinylidene difluoride membranes, and their N-terminal sequences were determined. The sequences of ChiP from V. furnissii and the E. coli recombinant proteins are compared with that predicted from the gene sequence (Table II).

Similarity of V. furnissii Chitoporin to Other Proteins-
In V. furnissii, the first 23 amino acids are processed, presumably as the protein enters the periplasmic space before insertion into the outer membrane. E. coli expresses both unprocessed (Fig. 5, arrow 1) and processed forms (Fig. 5, arrow 2) of the ChiP. Thus, the V. furnissii signal sequence can be recognized by the appropriate E. coli secretory pathway. Incomplete processing may have resulted from overexpression of the gene, which exceeded the capacity of the secretion/processing system of the E. coli cells. Indeed, if cells are grown at a slightly lower rate by decreasing growth temperature from 37 to 30°C, the processed form of ChiP increases from approximately 10 -20 to 50 -60% of the total ChiP expressed. Furthermore, it is possible that the outer membrane preparations are contaminated with inclusion bodies containing the unprocessed protein.
Construction of chiP Null Mutant in V. furnissii-The nucleotide sequence of the chiP gene was used to construct a null mutant in wild type V. furnissii as described under "Experimental Procedures." Briefly, a 0.3-kb fragment of the chiP gene was subloned into the suicide vector pNQ705 containing a chloramphenicol resistance selection marker. The resulting plasmid was transconjugated from the recombinant E. coli 5 Preliminary sequence data were obtained from The Institute for Genomic Research. The V. cholera sequence displayed the highest homology (49% indentity/60% similarity) to the V. furnissii chitoporin. We believe that the V. cholera protein is a chitoporin homologue.

FIG. 2. Separation of V. furnissii inner and outer membranes. V. furnissii
cells were grown to an absorbance at 540 nm ϭ 0.5 in lactate-ASW medium containing 0.6 mM (GlcNAc) 2 . Inner and outer membranes were separated using isopyncnic sucrose gradient centrifugation as described under "Experimental Procedures." A total of 25 fractions (1.0 ml/fraction) were isolated. Aliquots (0.1 ml) of each fraction were analyzed for KDO, II Nag activity, and for protein (؉) as described. q, outer membrane marker; E, inner membrane marker. The II Nag activity is expressed as nmol [ 14 C]GlcNAc-6-P formed per min per 0.1 ml of fraction. strain S17-1 in which it can replicate and where it was appropriately methylated, so that it was protected against the V. furnissii restriction system. Selection on chloramphenicol gave clones containing a knock-out of the chiP gene by homologous recombination. A single insertion in the V. furnissii genome corresponding to the region containing appropriate restriction sites for chiP was confirmed by Southern hybridization (data not shown).
No protein band corresponding to ChiP was observed in the null mutant under any conditions tested. The null mutant strain was complemented by a construct containing the chitoporin gene fused to a Ptac promoter (pSF-chiP) as described under "Experimental Procedures." Expression of the chitoporin in the complemented strain, designated V. furnissii XC1, was constitutive.
Growth of V. furnissii Strains on Chitin Oligosaccharides-Wild type V. furnissii, strain X1401 (chiPmutant), and strain XC1 (chiP ϩ complemented mutant) were grown at 30°C in minimal media containing one of the following carbon sources: 40 mM lactate, 20 mM GlcNAc, 10 mM (GlcNAc) 2 , or 6.7 mM (GlcNAc) 3 . The different sugar concentrations yield about the same number of GlcNAc equivalents in each mixture. The growth curves are presented in Fig. 6. The data show that the three cell types grow at the same rates on lactate, GlcNAc, and (GlcNAc) 2. By sharp contrast, the wild type and XC1 strains readily grow on the trisaccharide, (GlcNAc) 3 , whereas the chiP mutant grows very slowly, if at all, on this oligosaccharide over the time course of the experiment.
Transport of Me-TCB by Wild Type and Mutant Strains-The V. furnissii (GlcNAc) 2 transport system has been characterized using a radioactive nonmetabolizable analogue [ 3 H]Me-TCB (13). The effect of the chiP null mutation on the initial rate of Me-TCB uptake was measured as a function of the external concentration of the substrate (Fig. 7A). Initial rates were calculated from transport experiments conducted between 7 and 21 s in buffered 50% ASW. Wild type V. furnissii exhibits a 6-fold greater rate of uptake than the mutant at low substrate concentrations. At high substrate concentrations, the two rates  converge, and by 50 M external [ 3 H]Me-TCB, the calculated rates for uptake between the wild type and mutant strains are identical. Thus, the porin has a significant effect on the uptake rate of the disaccharide only at low concentrations (see "Discussion").
Regulation of chiP Expression-Cultures were grown on lactate or on various sugars including GlcNAc, galactose, glycerol, mannose, mannitol, maltose, sucrose, fructose, and glucose (all at 0.5%), with and without 0.6 mM (GlcNAc) 2 as inducer. ChiP expression as well as ␤-GlcNAcidase activities were quantified as described under "Experimental Procedures," and the data are presented in Table III. ␤-GlcNAcidase activity was measured in the detergent solubilized fraction; this value represents the sum of all the cellular ␤-GlcNAcidase activities, including both cytoplasmic and periplasmic enzymes (8,11,12,14). The genes encoding these enzymes do not appear to be in a single operon, and they are probably differentially regulated.
All the sugars tested, except perhaps galactose and glycerol, repressed both ChiP and ␤-GlcNAcidase induction. Glc seemed to be the most potent inhibitor, and it is interesting to note that the monosaccharide, GlcNAc, also inhibits induction of the "disaccharide" pathway.
To determine whether glucose repression of ChiP and ␤-Gl-cNAcidase expression was mediated via reduction in cAMP levels, cultures were grown in lactate or glucose medium with and without the inducer 0.6 mM (GlcNAc) 2 and with and without 10 mM cAMP. Table III shows that 10 mM cAMP alleviated glucose repression of expression of both ChiP and the ␤-Glc-NAcidases. As indicated in Fig. 4, the nucleotide sequence of the DNA upstream of chiP shows several potential cAMP/crp binding sites, one or more of which may be involved in this regulation of expression in V. furnissii.

DISCUSSION
The problem posed in the Introduction to this paper is how chitin oligosaccharides penetrate the cell envelope of V. furnissii so that they can be hydrolyzed by the specific periplasmic ␤-N-acetylglucosaminidases to the mono-and disaccharides. We speculated that a specific porin, or chitoporin, might be required, and in this paper we present evidence for such a porin, ChiP, encoded by the gene chiP.
Generally, constitutive or nonspecific porins are trimeric outer membrane proteins with pore sizes that permit entry of di-and sometimes trisaccharides, depending on the shapes of the molecules. There are a few sugar-specific porins, or glycoporins, the best characterized being LamB (20, 22, 40 -42), which permits the entry of high molecular weight glucose oligosaccharides. E. coli also expresses the specific ScrY porin (23), induced by sucrose.
RafY is induced by the trisaccharide raffinose (24), and Ulmke et al. make the following points: (a) Although the trisaccharide raffinose can diffuse through the constitutive, nonspecific E. coli porins (PhoE, OmpC, and OmpF), the diffusion rate is too slow to permit growth. The K m for raffinose uptake is 2 mM in the absence of RafY and 0.13 mM when it is expressed. RafY has a relatively broad specificity, which is explained by the fact that the reconstituted porin showed no specific binding of the trisaccharide (25), and the aqueous channel was wider than the constitutive E. coli porins. (b) The uptake rates of disaccharides such as maltose and sucrose are dependent on their concentrations. At 5 mM, the diffusion rates through the general porins are so rapid that they do not affect the rates of uptake (which are determined by the respective permeases); at these concentrations, there is no increase in uptake rate when the cells express glycoporins. By contrast, at low substrate concentrations, uptake rates of the disaccharides are increased when glycoporins are expressed. (c) Finally, increased uptake rates of the disaccharides are only observed when the porins are coupled to high affinity permeases (K m , 1-10 M) such as the maltose permease. In other words, under these conditions, diffusion through the porins is the rate-limiting step in uptake.
ChiP is believed to be a V. furnissii chitin oligosaccharide porin for the following reasons: (a) The protein is induced only by chitin oligosaccharides and is found in the outer membranes of V. furnissii. Induction is catabolite repressed and can be alleviated by the addition of cAMP. In addition, the recombinant protein is detectable in the outer membranes of E. coli. (b) The gene, chiP, that encodes the protein was cloned and expressed in E. coli. Although bacterial porins generally show little or no amino acid sequence similarity, the deduced amino acid sequence of ChiP shared homology with bacterial porins from several Neisseria species, B. pertussis, and with the Sal-monella and E. coli phoE porins. The cloned gene was used to construct a V. furnissii chiP Ϫ null mutant by using homologous recombination with a suicide vector. The cloned gene complements the mutant in the experiments described below. (c) V. furnissii expresses a (GlcNAc) 2 -inducible permease (13), and we have characterized the kinetics of this permease with a nonmetabolizable analogue of the disaccharide, Me-TCB. In the present studies, the radiolabeled disaccharide analogue was again used in transport experiments. At low concentrations (0.25-2 M), there is a marked difference in the kinetics exhibited by the wild type and chiP Ϫ mutant cells (Fig. 7), about 6-fold in the rate. As the concentration of the solute is increased Ͼ2 M, the slopes of the lines (wild type and mutant) become equivalent, and at "saturation" of the transporter (Ͼ50 M) the rates are equal. In our earlier work on this permease (13), we discussed the difficulties in obtaining valid kinetic data. With wild type V. furnissii cells, we obtained the following kinetic constants: apparent K m , 1 M (more accurately, Յ 1 M); apparent V max , 2.1 nmol uptake/min/mg cell protein (more accurately, Ն2.1 nmol/min/mg). Considering the technical problems, the present results are consistent with the earlier data (Fig. 7). In other words, as the level of the solute is increased, the permease becomes dominant in the uptake process, and it is only at low substrate concentrations that the effect of the porin is apparent. These results are therefore in complete accord with those reported for RafY discussed above. (d) Growth experiments (Fig. 6) were conducted with 20 mM Glc-NAc, and equivalent (in terms of carbon content) concentrations of the di-and trisaccharide, as well as lactate. The wild type, chiP Ϫ mutant, and the transformant (complemented chiP Ϫ ) cells grow at essentially the same rates on lactate, GlcNAc, and (GlcNAc) 2 . There was, however, a marked difference in their growth properties on 6.7 mM (GlcNAc) 3 . Whereas the wild type cells and the complemented mutant grew on the trimer at about the same rate as on the other carbon sources, the chiP Ϫ null mutant showed little to no growth on the trisaccharide over the time course of the experiment. 6 Taken together, these results provide substantial evidence that the V. furnissii outer membrane protein is a chitoporin. Whether it contains a specific chitin oligosaccharide binding site such as LamB or is simply a wider channel as appears to be true of RafY remains to be established. Current studies are aimed at more precisely defining the solute properties of ChiP in reconstituted vesicles. a The minimal medium was buffered 50% ASW supplemented with the indicated carbon source at 10 mM concentration. The compounds above the space are not substrates of the PTS in this organism, whereas those below the space are substrates. PTS substrates, such as glucose, are strong catabolite repressors in E. coli and other bacteria.
b ChiP was determined as described in the text (Coomassie-stained SDS-PAGE gels) and is presented as the percentage of total outer membrane protein. ND, not detected, less than 1% of the total protein in the lane. c ␤-GlcNAcidase activity in crude extracts was determined by measuring the rate of hydrolysis of p-nitrophenyl ␤-N-acetyl-D-glucosaminide; activity ϭ nmol/min/g protein at 25°C. It should be emphasized that the extracts contain several enzymes that hydrolyze this substrate. d Higher chitin oligosaccharides, (GlcNAc) n , n ϭ 3-6, gave essentially the same results as (GlcNAc) 2 . Glucosamine, its oligosaccharides, and cellobiose did not induce expression of chiP. e A protein band that migrated at the same rate as ChiP was observed in membranes from cells grown on GlcNAc as the sole carbon source, but it is not known whether the GlcNAc-induced protein is the same as ChiP.