Isolation of Cyanophycin-degrading Bacteria, Cloning and Characterization of an Extracellular Cyanophycinase Gene ( cphE ) from Pseudomonas anguilliseptica Strain BI THE cphE GENE FROM P. ANGUILLISEPTICA BI ENCODES A CYANOPHYCIN-HYDROLYZING ENZYME*

Eleven bacteria capable of utilizing cyanophycin (cy-anophycin granule polypeptide (CGP)) as a carbon source for growth were isolated. One isolate was taxonomically affiliated as Pseudomonas anguilliseptica strain BI, and the extracellular cyanophycinase (CphE) was studied because utilization of cyanophycin as a carbon source and extracellular cyanophycinases were hitherto not described. CphE was detected in supernatants of CGP cultures and purified from a corresponding culture of strain BI employing chromatography on the anion exchange matrix Q-Sepharose and on an arginine-agarose affinity matrix. The mature form of the inducible enzyme consisted of one type of subunit with M r (cid:1) 43,000 and exhibited high specificity for CGP, whereas proteins and synthetic polyaspartic acid were not hydrolyzed or were only marginally hydrolyzed. Degradation products of the enzyme reaction were identified as aspartic acid-arginine dipeptides under vigorous shaking. For heat inactivation of the enzyme, 50- (cid:3) l samples were transferred to test tubes con- taining 500 (cid:3) l of water preincubated at 70 °C. After 5 min of inactivation, the samples were transferred onto ice for 5 min to allow for CGP reprecipitation. After subsequent centrifugation, 50 (cid:3) l of supernatant were mixed with 500 (cid:3) l of Hydroluma ® Scintillation mixture (J. T. Baker, Inc.). Radioactivity was measured with a model LS 6500 scintillation counter (Beckman Instruments (27)).

most cyanobacteria as nitrogen, carbon, and energy storage compound in the early stationary growth phase (1,2). The water-insoluble CGP is accumulated intracellularly in the form of membraneless granules (3) and is degraded by the cells when growth is resumed. The backbone of this unique biopolymer consists of ␣-amino-␣-carboxyl-linked L-aspartic acid monomers. Most of the ␤-carboxylic groups are covalently bound to the ␣-amino groups of L-arginine residues (4,5); in recombinant Escherichia coli expressing cyanobacterial CGP-synthesizing enzymes (see below), a significant fraction of arginine is replaced by lysine (6).
Although much information has been obtained concerning the non-ribosomal biosynthesis of CGP, which is catalyzed by the cyanophycin synthetase (CphA; see Ref. 7 and cited references therein), only a few reports are available on the intracellular degradation of CGP. Intracellular CGP degradation was first observed in crude extracts of soluble proteins prepared from cells of Anabaena cylindrica (5). The corresponding enzyme, cyanophycinase (CphB), was purified from a recombinant E. coli harboring the cphB gene from Synechocystis sp. PCC6803 and characterized in detail (8). Dipeptides consisting of arginine plus aspartic acid and free arginine were identified as products of CGP degradation in addition to small amounts of aspartic acid (8). In contrast to intracellular degradation, nothing is known about the extracellular decomposition of this biopolymer by bacteria or other microorganisms.
In this study, we demonstrate for the first time that CGP can be easily degraded and utilized as the sole carbon source for growth by a variety of non-cyanobacterial eubacteria isolated from different habitats. Because it is known that CGP is resistant to a wide range of commercially available proteases (4,9), these bacteria must possess an enzyme specialized for CGP degradation. We report on the isolation of a strain of the species Pseudomonas anguilliseptica and describe the substrate utilization capabilities of this bacterium and the purification of an extracellular cyanophycinase (extracellular CGPase (CphE)) from culture supernatants of cells grown on CGP. Furthermore, CphE was biochemically characterized to reveal the degradation mechanism and to identify the cleavage products. In addition, the CGPase gene (cphE) of the isolated P. anguilliseptica strain BI was cloned and characterized.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth of Bacteria-CGP-degrading bacteria isolated in this study are listed in Table I. These strains were either grown on Standard I complex medium (Merck) or grown on basic inorganic medium B (10) for CGP degradation and substrate utilization experiments. The concentrations of CGP and other carbon sources added to the medium are indicated in the text. All isolates were grown at 30°C. The following microorganisms were used as reference strains in substrate utilization assays on solid CGP medium (see below) with 0.05% (w/v) glucose as an additional carbon source: E. coli K12 (wild type), Pseudomonas putida KT2440 (11), Micrococcus luteus (DSMZ 348), Bacillus subtilis 168 ϩ (DSMZ 402), and Bacillus megaterium (DSMZ 319). For CGP production, an E. coli DH1 strain harboring plasmid pMa/c5-914::cphA expressing cphA from Synechocystis PCC6803 (12) was employed (see below). E. coli strains were usually grown at 37°C in Luria-Bertani (LB) medium or terrific broth (TB) complex medium (13).
Preparation of Selective Medium for the Detection and Isolation of CGP-degrading Microorganisms-Samples from different sources were spread on solid basic inorganic medium B (10) supplemented with trace element solution SL 7 (14) and overlaid with 0.5% (w/v) agar containing 0.2% (w/v) of CGP. For the preparation of the overlay agar, diethyl ether-sterilized CGP was first dissolved in 0.1 N HCl and then added to sterile medium under vigorous stirring to avoid the formation of inhomogeneous CGP precipitates. For adjustment of the pH value, an equal volume of 0.1 N NaOH was added before pouring plates.
Isolation, Manipulation, and Analysis of DNA-For isolation of plasmid DNA, the lithium preparation method was applied (15). Total genomic DNA of P. anguilliseptica strain BI was isolated according to the method of Rao et al. (16). After partial digestion with the endonuclease PstI, genomic DNA fragments were ligated to the cosmid vector pHC79 (17), and E. coli strain S17-1 (18) was used as a recipient for transduction of the cosmid library. A Gigapack ® III XL packaging extract (Stratagene, La Jolla, CA) was employed for the packaging of DNA and subsequent infection of strain S17-1 as described by the manufacturer. E. coli strain XL1-Blue (Stratagene) was used in combination with pBluescript SK Ϫ (Stratagene) for cloning of a PstI restriction subfragment (2600 bp), sequence analysis of cphE, and heterologous production of the enzyme.
The 16-S rRNA gene was amplified from total DNA (see above) using oligonucleotide primers as described before (22). After purification of the PCR products with a NucleoTrap ® CR kit (Macherey-Nagel, Dü ren, Germany), their nucleotide sequences were determined as described above. The 16-S rDNA sequence was aligned with published sequences from representative Pseudomonas species from the National Center for Biotechnology Information (NCBI) data base.
Nucleotide Sequence Accession Numbers-The nucleotide and amino acid sequence data reported here for cphE have been submitted to the NCBI data base under accession number AY065671. The 16-S rRNA gene sequence data of P. anguilliseptica strain BI were deposited in the NCBI data base under accession number AF439803.
Cyanophycin Production-For production of native CGP, Synechocystis sp. strain PCC6308 was cultivated in full-strength BG11 medium (23) in an 80-liter closed tubular glass photobioreactor as described before (24). Also, a recombinant E. coli DH1 harboring plasmid pMa/c5-914::cphA (see above) with a temperature-sensitive inducible promoter was employed for the production of CGP. A 42-liter Biostat UD30 stainless steel bioreactor (B. Braun Biotech International, Melsungen, Germany) with TB complex medium was used for production as described previously (12).
Isolation of Cyanophycin-After cell harvest, CGP was isolated according to the method of Simon (25), which was modified by applying only one washing step at each Triton X-100 ® concentration and two additional centrifugation steps of the acidic and neutralized suspensions, respectively. The purity of CGP was controlled both by SDS-PAGE with subsequent Coomassie staining (26) and by HPLC analysis after acid hydrolysis of the polymer and subsequent derivatization of amino groups with o-phtaldialdehyde reagent (27).
Taxonomic Determination with Physiological Tests-Motility and Gram behavior were determined as described before. Oxidase (Bactident Oxidase test strips from Merck) and catalase tests were performed according to standard protocols. Further determinations were done by using the API 20NE test kit (BioMérieux, Marcy-l'Etoile, France). Analytical Methods-Reversed phase HPLC was used to determine the products of enzymatic CGP degradation as described by the method for the quantitative determination of amino acids (27). Electrospray ionization mass spectrometry (ESI) was applied for identification of the final degradation product of CGP by mass determination and structural analysis (28). All measurements were performed employing a Quattro LCZ system (Micromass, Manchester, UK) with a nanospray inlet.
Purification of the Extracellular Cyanophycinase from P. anguilliseptica Strain BI-A cell-free supernatant from a CGP culture was obtained by sedimentation of the cells in the late exponential growth phase by centrifugation and subsequent filtration of the supernatant through a 0.2-m nitrocellulose membrane. All steps were carried out at 4°C and in the presence of 50 mM sodium phosphate buffer (pH 8.3). Further components added to the buffer are mentioned below. After concentration in an ultrafiltration chamber (Amicon, Beverly, MA) using a YM10 membrane, the buffered solution was applied onto a MonoQ HR5/5 anion exchange column (Amersham Biosciences). After washing the column with 2 bed volumes of buffer, CGPase was eluted with a linear NaCl gradient (0 -1 M) employing an increase of NaCl concentration of 17 mM/ml and a total flow rate of 1 ml/min. Active fractions (1 ml) were detected after the transfer of 10 l of the respective eluates onto CGP overlay plates (see above) by the occurrence of halos after 5-40 min of incubation at 30°C. Fractions with high activity were combined, desalted by ultrafiltration (see above), and applied onto an arginineagarose column (5-ml bed volume; Sigma). For selective elution of the enzyme, an arginine gradient (0 -1 M) was applied. To avoid nonspecific protein binding and to prevent the enzyme from binding in the presence of high arginine concentrations, the buffer in addition contained 100 mM NaCl.
SDS-PAGE of active enzyme fractions or CGP samples was performed in 11.5% polyacrylamide gels according to standard protocols (29). Proteins were visualized by the Coomassie staining method (26). An "in-gel" renaturation method described for activity staining with proteases after SDS-PAGE (30) was used to obtain reactivated CGPase after separation of enzyme subunits according to their apparent molecular mass under denaturating conditions. The ability of reactivated CGPase to form degradation halos was tested by the application of a thin CGP-agar layer (see above) on top of buffer-pretreated gels. Protein concentrations were determined by the procedure of Bradford (31).
Characterization of the Purified CGPase-For determination of the substrate specificity of the CGPase, the purified recombinant enzyme was incubated at 30°C in 1 ml of 50 mM sodium phosphate buffer (pH 8.3) with various polypeptide substrates. Each reaction contained 1 mg of the respective substrate and 1.6 g of enzyme. The reaction was stopped after 120 min by incubation at 70°C for 5 min. After centrifugation, 100-l aliquots of supernatant were incubated at 95°C for 5 min in the presence of 1.25% ninhydrin (Merck) in 1 ml of total reaction volume. Subsequently, they were assayed photometrically at 570 nm for the presence of released hydrolysis products. Bovine casein (Hammersten grade) was from Merck, bovine serum albumin (BSA) was from Roth (Karlsruhe, Germany), and poly(␣,␤-D/L-aspartic acid) (M r ϭ 11,000) was obtained from Bayer (Leverkusen, Germany). Labeling experiments were performed by enzymatic elongation of the C terminus of a CGP primer (32). L-[U-14 C]arginine was incorporated into the polymer chain using purified cyanophycin synthetase from Synechocystis sp. strain PCC6308 heterologously produced in E. coli (27). Labeled CGP was incubated at 30°C with 1.6 g of CGPase in 50 mM sodium phosphate buffer (pH 8.3) under vigorous shaking. For heat inactivation of the enzyme, 50-l samples were transferred to test tubes containing 500 l of water preincubated at 70°C. After 5 min of inactivation, the samples were transferred onto ice for 5 min to allow for CGP reprecipitation. After subsequent centrifugation, 50 l of supernatant were mixed with 500 l of Hydroluma ® Scintillation mixture (J. T. Baker, Inc.). Radioactivity was measured with a model LS 6500 scintillation counter (Beckman Instruments (27)).

Enrichment and Isolation of CGP-degrading Bacteria-To
screen for CGP-degrading bacteria, samples from typical habitats of CGP-producing cyanobacteria were plated onto solid mineral medium containing CGP as the sole carbon source. Due to the insolubility of CGP at neutral pH, the agar was turbid. Colonies of CGP-degrading microorganisms were recognized because of the formation of degradation halos, which appeared after 12-18 h of incubation at 30°C (Fig. 1A). Based on this feature, axenic cultures of nine bacterial strains were finally isolated from Baltic sea water, different pond sediments, and sewage sludge (Table I). In addition to the newly isolated CGP-degrading strains, other bacteria from our culture collection were also tested, and two additional strains with CGP degradation capability were detected (strains BE2 and PAS1, Table I). However, E. coli K12, P. putida KT2440, M. luteus, B. subtilis 168 ϩ , and B. megaterium were not able to cause formation of halos on CGP overlay agar plates, although some of these bacteria (e.g. B. subtilis) are known to use proteins as nutrients.
Characterization of CGP-degrading Bacteria through Substrate Utilization Patterns and Taxonomic Classification-All isolates tested (from Baltic sea water, pond sediment, and sewage sludge, i.e. "A, B, and D series"; Table I) were Gramnegative, oxidase-and catalase-positive rod-shaped bacteria. With the exception of isolate DXIII, all strains showed motility. Applying the API 20NE test kit, two isolates (DIII and DIV) revealed acceptable identification profiles. Both strains were taxonomically affiliated as strains of the species Pseudomonas alcaligenes (Table II). As listed in Table I, most strains isolated in this study showed growth on the amino acid constituents of CGP, i.e. on aspartic acid and arginine. For most strains, growth with arginine was faster than with aspartic acid. Only isolate BI showed no growth on aspartic acid. Strain PAS 1 from the culture collection of our institute also did not grow on aspartic acid as the sole carbon source. None of the isolates was able to grow on synthetic poly(␣,␤-D/L-aspartic acid) (data not shown). With poly(␥-D-glutamic acid) as the sole carbon source, only isolate DXIII and to some extent also P. alcaligenes strain BE2 (33) showed growth (Table I). With bovine serum albumin as the sole carbon source, only strain PAS I and P. alcaligenes strain BE2 exhibited good or poor growth, respectively. Citrulline and ornithine, two putative degradation products of arginine, were not utilized as carbon sources for growth by any of the bacteria investigated in this study. The only exceptions were isolate AV N , which was able to grow on ornithine, and B. subtilis 168 ϩ , which utilized citrulline (data not shown).
Taxonomic Classification of Isolate BI by 16-S rRNA Gene (rDNA) Sequence Analysis-For several reasons, isolate BI was a good candidate for a more detailed investigation of CGP degradation. Therefore, the taxonomic position of the isolate was determined. Analysis of the 16-S rDNA sequence of isolate BI revealed 98% identity to the nucleotide sequence of all three P. anguilliseptica strains available at the NCBI data base including the P. anguilliseptica type strain NCIMB 1949. Maximum sequence identity to other species of the genus Pseudomonas was only in the range of 95-96% (Fig. 2). Therefore, the new isolate was referred to as P. anguilliseptica strain BI.
Preliminary 16-S rDNA sequence data of strain PAS 1 (about 1000 bp) revealed that this strain most probably belongs to the genus Streptomyces. This finding corresponds well with the streptomycete-like habitus of this strain, e.g. the formation of exospores in aging colonies.
Growth Kinetics of P. anguilliseptica Strain BI-The ability of P. anguilliseptica strain BI to grow on CGP as the sole carbon source was investigated in more detail. Therefore, growth of this strain on CGP and on its amino acid constituents as well as on the non-related substrate citrate was monitored over 24 h (Fig. 3). Living cell counts for the cyanophycin culture revealed that growth of the cells started at about 4 h of incubation (data not shown) after inoculation from a citrate culture. The turbidity caused by suspended CGP particles disappeared visibly during incubation. Strain BI grew best with a combination of arginine and aspartic acid if these amino acids were provided at a molar ratio according to their proportional masses in the CGP molecule (248 Klett units). Growth on CGP led to a maximum optical density of 202 Klett units, which is in the range of the OD of the citrate culture. Slightly weaker growth was detected for the arginine culture (182 Klett units).
No increase or change of the OD occurred in the control (sterile medium containing citrate) or in mineral salt medium containing aspartic acid as the sole carbon source (Fig. 3). During growth on CGP, 46% (w/w) of the polymer was converted into cellular dry matter by P. anguilliseptica strain BI.
Purification of the Extracellular CGPase from P. anguilliseptica Strain BI-The extracellular CGPase of P. anguilliseptica was purified to electrophoretic homogeneity from CGP-grown cultures by the application of anion exchange chromatography on Q-Sepharose followed by L-arginine-agarose affinity chromatography (Fig. 1C). The latter is usually used for different purposes, e.g. purification of transfer RNA molecules (34). The L-arginine-agarose matrix was highly specific for the binding of CGPase under the employed conditions, revealing a high affinity of the enzyme to this matrix. Therefore, an arginine gradient (0 -1 M) in sodium phosphate buffer was applied for the elution of the CGPase. To further reduce nonspecific binding of other proteins, the sodium phosphate buffer additionally contained 100 mM NaCl. In the absence of NaCl, arginine had to be applied at concentrations higher than 3 M for total release of the enzyme from the affinity matrix, again reflecting the high affinity of the CGPase to the arginine residues coupled to the matrix. During enzyme purification, active fractions were detected by their ability to cause rapid halo formation on CGP overlay agar plates (Fig. 1B). These halos occurred mostly within a few minutes to 2 h, and their diameters indicated the activity of the enzyme in the sample.
SDS-polyacrylamide gel electrophoresis revealed an apparent molecular mass of the subunits of the enzyme of 43 kDa (Fig. 1C). By the employment of an in-gel renaturation method, previously used for the detection of proteases in SDS-polyacrylamide gels (30), in combination with the subsequent application of a thin CGP-agar layer on top of the "renaturated" gel, it was possible to detect reactivated CGPase by the formation of a degradation halo at the position corresponding to a protein of the expected molecular mass of 43 kDa (Fig. 1D). This finding clearly proved that the 43-kDa protein represented the subunit of the CGPase and that the enzyme consisted of only one type of subunit or of subunits with identical apparent molecular masses.
Purification of the P. anguilliseptica CGPase from a Recombinant Strain of E. coli-The heterologously produced enzyme was purified in the same way from the soluble fraction of E. coli cells harboring pBluescript SK Ϫ ::cphE (for construction, see below) grown in TB medium. The enzyme was not excreted from the cells when it was produced by E. coli. In culture supernatants, enzyme activity was not detectable. In addition, halo formation on CGP overlay LB agar plates occurred only after 3 days of incubation, thus indicating that the release of the enzyme presumably occurred due to partial cell lysis during aging of the cells. By this purification method, the activity of the heterologously expressed enzyme was enriched 15-fold, and 49% of the total activity was recovered as confirmed by the calculation of halo forming units in a plate diffusion assay using purified enzyme as a standard (data not shown).
Cloning and Analysis of the Extracellular CGPase Gene from P. anguilliseptica BI-Applying the restriction endonuclease PstI for partial digestion of genomic DNA from P. anguilliseptica BI, fragments with a broad size range were obtained and subsequently ligated to the cosmid vector pHC79. After transduction of the cosmid library into E. coli S17-1, ϳ3000 tetracycline-resistant clones were obtained and tested for their ability to degrade CGP by transfer onto LB agar plates overlaid with a thin layer of CGP-containing medium. After 3 days of incubation at 37°C, one colony was detected that was surrounded by a halo, indicating degradation of the polymer in the CGP layer. The plasmid containing the CGPase-encoding genomic fragment from P. anguilliseptica BI (26 kbp) was isolated from the respective clone. Twelve subfragments were obtained after total digestion with PstI and ligated to the vector pBluescript SK Ϫ . Subsequent transformation of E. coli XL1-Blue with the resulting mixture of pBluescript SK Ϫ construction products led to the identification of a bacterial colony capable of forming halos on a CGP overlay plate.
Biochemical Characterization of the CGPase-P. anguilliseptica strain BI produced the extracellular CGPase only when CGP was present in the medium. In complex Standard I medium, no enzyme activity was detected. As substrate utilization patterns of strain BI indicated (Table I), no correlation occurred between the ability to degrade CGP and the utilization of other polyamide substrates, including BSA, poly(␥-D-glutamic acid), or synthetic poly(␣,␤-D/L-aspartic acid) (data not shown). This indicated that the CGPase is not employed by the bacterium for nonspecific hydrolysis of polyamide substrates. This was also confirmed by studies on the substrate specificity of the purified enzyme. Using the enzyme purified from the recombinant strain of E. coli (see above), the release of degradation products from the polyamide substrates CGP, BSA, bovine casein, and poly(␣,␤-D/L-aspartic acid) was investigated by employing ninhydrin reagent for detection of released amino groups (Fig. 4). After 2 h of incubation in the presence of purified enzyme, only CGP samples showed a significant release of ninhydrin-positive degradation product, whereas BSA and bovine casein samples revealed only a very weak release of reactive products amounting to 3.9 or 6.4% of that obtained with CGP, respectively (Fig. 4). From synthetic poly(␣,␤-D/L-aspartic acid), no release of ninhydrin-positive material was detected (Fig. 4).
Growth was qualitatively estimated as follows: -, no growth; ϩ, poor growth; ϩϩ, moderate growth; ϩϩϩ, good growth; PGA, poly(␥-D-glutamic acid). Strains were grown on basic inorganic medium B with concentrations of 0.25% of each of the applied carbon sources.
c Strain BE2 was isolated in a screening for poly(␥-D-glutamic acid) degrading bacteria (33).  Using purified recombinant CGPase, degradation of cyanobacterial CGP (Synechocystis PCC6308) was visualized by SDS-PAGE and subsequent nonspecific protein staining (Fig.  5). Degradation of the high molecular weight polydisperse biopolymer (about 43-100 kDa; lane 2) to low molecular weight material (lanes 3-9) is demonstrated in Fig. 5. The lack of detectable high molecular weight material after 165 min of incubation (lane 10) corresponded well with the nearly exclusive detection of the final degradation product of CGP (see HPLC and ESI analysis below), indicating total degradation of CGP having occurred within that time. A splitting of the initial molecule population (lane 2) into two populations of molecules exhibiting two different molecular weight ranges was observed during the incubation (Fig. 5, lanes 3-9).
The composition of the amino acid constituents of CGPs varies depending on the source of CGP. Cyanobacterial CGP isolated from Synechocystis PCC6308 contains only aspartic acid plus arginine, whereas in CGP isolated from cells of a recombinant E. coli expressing the PCC6803 cyanophycin synthetase gene, some arginine residues were replaced by lysine residues (6). The final degradation products of the enzyme reaction on both CGPs were determined. As shown in Fig. 6, degradation of the cyanobacterial CGP led to the formation of only one detectable degradation product after separation in HPLC, whereas recombinant polymer produced by E. coli led to the formation of two main products. Therefore, it seemed likely that dipeptides of ␤-Asp-Arg or of ␤-Asp-Arg plus ␤-Asp-Lys, respectively, were formed during the degradation process. A degradation mechanism producing oligomers would have led to the formation of more than two different products in the case of the recombinant CGP, resulting most probably in the appearance of more than two peaks in the HPLC chromatogram. Final proof for the presence of the dipeptide was obtained by the
To elucidate the mechanism of CGP degradation, C-terminal L-[U- 14 C]arginyl-labeled CGP was synthesized using the cyanophycin synthetase from Synechocystis PCC6308. Incubation of labeled CGP with recombinant CGPase resulted in an immediate release of radioactivity that continued for ϳ7 min (Fig.  8). Together with the observation that after addition of CG-Pase, the release of ␤-Asp-Arg dipeptides was detectable by HPLC analysis within 30 s (data not shown), an exo-degradation mechanism proceeding (at least partially) from the C terminus of CGP seems to be most likely for the enzyme reaction.
Inhibitor studies showed that the extracellular CGPase of P. anguilliseptica BI is strongly inhibited by the serine protease inhibitors phenylmethylsulfonyl fluoride and Pefabloc ® (63 and 92%, respectively; Table III). The inhibition of the enzyme by Pefabloc ® was additionally confirmed by the inhibition of halo formation (Table III). CGPase activity was only slightly (and inhibitor concentration independently) decreased by the thiol protease inhibitor leupeptin (13%). The observed decrease in activity in this case is most probably due to interference with the "o-phtaldialdehyde-derivatization" method necessary for HPLC analysis. The application of the metalloprotease inhibitor EDTA led to a strong reduction of the release of detectable ␤-Asp-Arg dipeptides in the enzyme reaction (Table III). This was most probably due to the formation of precipitates occurring during derivatization. The ability of the enzyme to form degradation halos in CGP overlay agar in the presence of 30 -500 mM EDTA was, however, not affected. Only the tryptophan oxidant N-bromosuccinimide totally prevented the release of ␤-Asp-Arg and the formation of degradation halos on CGP overlay plates. The latter occurred after 30 min of incubation together with CGPase, even if N-bromosuccinimide was applied at concentrations Ն1 mM. Therefore, a tryptophan residue may be involved in the process of CGP degradation (compare Fig. 9).
Molecular Characterization of the Extracellular CGPase Gene (CphE) from P. anguilliseptica-The coding gene for CphE from P. anguilliseptica strain BI was identified by its heterologous active expression from a gene library of total DNA in E. coli S17-1 (see above). One E. coli clone harboring a 26-kbp fragment of P. anguilliseptica genomic DNA exhibited the ability to cause the formation of degradation halos after 3 days of incubation on CGP overlay agar plates. After subcloning of smaller DNA fragments in pBluescript SK Ϫ , one transformant of E. coli XL1-Blue was identified that was also able to form degradation halos on CGP plates. This clone harbored a 2.6-kbp fragment of P. anguilliseptica DNA. By DNA sequence analysis of the cloned fragment, the N terminus of CphE, which was determined by N-terminal amino acid sequence analysis (compare Fig. 9), was rediscovered in antilinear orientation toward the lacZ promoter of the vector; the gene can therefore be assumed to be under the control of its own promoter. Upstream of the N terminus of the mature enzyme, which was determined by N-terminal sequencing, a probable leader peptide of 21 amino acids was found in the deduced amino acid sequence of the gene (Figs. 9 and 10). 9 -14 base pairs upstream of the methionine codon (ATG) of the leader peptide sequence, a purine-rich sequence GGAGAA was detected, indicating a potential ribosome binding site (Shine-Dalgarno-sequence) in the complementary mRNA transcript of the gene. An open reading frame of 1,254 bp with TAA as stop codon was identified corresponding to a theoretical protein mass of 42.4 kDa for the mature CphE protein if the mass of the leader peptide (2.4 kDa) is not considered. This corresponds well with the apparent molecular mass of the enzyme (subunit) that was detected by SDS-PAGE (43 kDa, compare Fig. 1). The pI of the mature enzyme (i.e. the extracellular form of CphE) was calculated to be 5.92.
Alignment of the deduced amino acid sequence of cphE with proteins exhibiting an acceptable sequence similarity (27-32% in conserved regions) revealed a 3-amino acid motif most probably representing the catalytical triad also present in other serine type proteases (Fig. 9). The differences between CphE, CphB enzymes, and the most closely related protein, a hypo-thetical protein of Caulobacter crescentus (Fig. 11), reveal the relatively isolated position for CphE among all known enzymes involved in CGP metabolism.

DISCUSSION
Employing a newly developed CGP mineral medium, CGPdegrading bacteria were isolated from different habitats where cyanobacteria and concomitantly CGP were expected to be present. Every sample applied to the CGP medium, regardless of the origin of the sample, led to the identification of Gramnegative bacteria capable of utilizing CGP as the sole carbon source for growth. This finding revealed the abundance of CGPdegrading eubacteria, especially of the genus Pseudomonas, to which three isolates (P. anguilliseptica BI, P. alcaligenes DIII and DIV) were taxonomically assigned in this study. In addition, strain PAS1, which most probably belongs to the genus Streptomyces, exhibited good growth on CGP, indicating that the capability to degrade CGP occurs also among Gram-positive bacteria. Occurrence of so many CGP-degrading bacteria is not surprising because cyanobacteria represent a large and metabolically highly diverse group of bacteria. Representatives of cyanobacteria occur in almost any aquatic and terrestrial environment exposed to light, and in some of these environments, cyanobacteria are the predominant microorganisms. Furthermore, most cyanobacteria are able to synthesize CGP. Moreover, the finding that CGP-like polymers are also synthesized by eubacteria not belonging to the group of cyanobacteria (35) suggests that such degradation mechanisms are even present in obligate heterotrophic microbial communities. In conclusion, CGP is probably an abundant biopolymer in natural environments. However, studies on the extracellular degradation of CGP were limited in the past due to the difficulties of producing sufficient amounts of this biopolymer by cultivation of cyanobacteria.
Because this enzyme, after CphA and CphB, is the third bacterial enzyme that is involved in CGP metabolism, and because it is localized extracellularly, the cyanophycinase of P. anguilliseptica enzyme was referred to as CphE. As demonstrated in this study, CphE of P. anguilliseptica BI was only synthesized if CGP was available as substrate for growth, indicating a specific induction of the enzyme by CGP or its degradation products. Moreover (Table I), it also became obvious that there was no correlation between the abilities to degrade CGP and to hydrolyze other polyamide substrates. Most of the other polyamide substrates were only poorly utilized by the employed strains. For example, the naturally occurring poly(␥-D-glutamic acid), which is an extracellular polymer of various Gram-positive bacteria (36 -39), was only accepted by isolate DXIII and by P. alcaligenes strain BE2, which was previously isolated on poly(␥-D-glutamic acid)-containing me-dium (33). On the other side, by cultivation of non CGPdegrading reference strains including typical protease producing bacteria (i.e. B. subtilis and B. megaterium) on CGP overlay agar plates, the resistance of the polymer against typical bac-TABLE III Inhibition of the CGPase by group-specific protease inhibitors CGPase from P. anguilliseptica BI was incubated in sodium phosphate buffer (pH 8.3) for 30 min at room temperature in the presence of the listed inhibitors and subsequently applied onto A) CGP-overlay agar plates or to B) suspended CGP. Inhibition of the enzyme was detected by delayed or total prevention of halo formation on CGP-overlay plates and by HPLC analysis (detection of the degradation product after OPA-derivatization, compare Fig. 6). The control was without inhibitor.  9. Alignment of deduced amino acid sequence of CGPase CphE from P. anguilliseptica strain BI with protein sequences of highest similarity (NCBI data base). A potential signaling peptide preceding the N terminus of CphE is underlined. The N terminus (shaded in light gray) was determined by peptide sequencing of purified native CGPase. The proposed residues of the catalytic triad are shaded in gray and are indicated by the symbols ,@ , and ƒ for serine, aspartic acid (or glutamic acid), and histidine. A potential alternative catalytic aspartic acid residue of CphE is shaded in light gray. The conserved Gly-Xaa-Ser-Xaa-Gly motif of serine type proteases is boxed. Three tryptophan residues whose oxidation could have lead to enzyme inactivation (compare Table III terial proteases, of which B. subtilis produces a great variety (40 -45), became clearly evident. This is consistent with the finding of the total resistance of CGP to a variety of commercially available proteases (4,9).
Studies on the substrate specificity of CphE of P. anguilliseptica BI showed that the enzyme is highly specific for CGP. Therefore, consistent with the findings mentioned above, CGP seems to be exclusively hydrolyzed by the employment of specialized CGPases. The lack of release of significant amounts of ninhydrin-positive degradation products from bovine casein, BSA, or synthetic poly(␣,␤-D/L-aspartic acid), which is used as a biodegradable substitute for non-degradable polyacrylates in detergents (46), during incubation in the presence of purified enzyme (Fig. 4) clearly demonstrated the high specificity of CphE for CGP. A similar high specificity toward CGP was also described for the intracellular enzyme (CphB) of Synechocystis sp. PCC6803 (8).
Purification of CphE from culture supernatants to electrophoretic homogeneity became very efficient by utilizing the high affinity of this CGPase toward immobilized arginine residues, thus allowing purification of CphE by the application of only two chromatography steps and with high activity yield.
The strong binding of the extracellular CGPase to immobilized arginine residues was not only observed during protein purification employing an arginine-agarose column. Also, the binding of CphE to the natural substrate CGP during cultivation occurred with such high affinity that soluble enzyme activity was detected in culture supernatants of CGP-degrading bacteria in liquid medium only after all CGP particles in the medium had visibly disappeared, and the enzyme was thereby released from the substrate. After cloning of the cphE gene from P. anguilliseptica BI and its heterologous functional expression in E. coli, sufficient amounts of the enzyme were obtained for further biochemical characterization. HPLC and ESI analysis identified dipeptides as degradation products of the enzyme reaction. In the case of cyanobacterial CGP, ␤-Asp-Arg dipeptides occurred (Figs. 6 and 7) as described for the intracellular cleavage of CGP in the cyanobacterium Synechocystis sp. PCC6803 (8). Degradation experiments employing enzymatically C-terminal L-[U-14 C]arginyl-labeled CGP revealed a continuous and immediate release of radioactive degradation product after addition of CGPase (Fig. 8). Because ␤-Asp-Arg dipeptides were detectable immediately after addition of the enzyme, an exo-degradation mechanism proceeding from the C terminus of the CGP molecule and release of the dipeptides by successive cleavage of the ␣-amide bonds of the polymer backbone seems to be most likely for CphE.
Molecular characterization of cphE revealed a DNA sequence that encodes a protein with a similarity of only 27-28% to intracellular CGPases (CphB) from cyanobacteria in conserved regions. In contrast to the intracellular CGPase from Synechocystis sp. PCC6803, which has an apparent molecular mass of 27 kDa (8), the molecular mass of extracellular CphE from P. anguilliseptica BI was significantly higher (43 kDa). The dendrogram shown in Fig. 11 demonstrates the isolated position of CphE.
The amino acids Ser 169 , Glu 185 , and His 222 of CphE may be the catalytic active residues responsible for the hydrolytic cleavage of the ␣-amide bonds of the polymer backbone (Fig. 9). Accordingly, the catalytic mechanism is suggested to be that of a serine type protease. This finding is in good agreement with the detected sensitivity of CphE toward serine protease inhibitors (compare Table III). In contrast to most serine type proteases, the characteristic aspartic acid residue of the catalytic triad is replaced by glutamic acid. The same amino acid replacement was observed for the intracellular CGPase of Synechocystis sp. PCC6803 or other cyanobacteria (see Ref. 8 and compare Fig. 5). In the predicted sequence of a hypothetical protein from C. crescentus, which showed the highest similarity to CphE in a NCBI data base search (32% identity over 325 amino acids) and exhibited a similar molecular mass, an aspartic acid residue typical for the catalytic triad of most serine proteases was present (Fig. 5). Therefore, Asp 188 of CphE, which is according to the alignment close to the position of the proposed catalytic glutamic acid residues of CphB proteins (8), must be considered as another potential residue that is involved in catalysis, instead of Glu 185 . The proposed catalytic aspartic acid residue of the PepE protein, which represents an aspartyl-dipeptidase from Salmonella typhimurium (47,48), has been suggested to be in a corresponding position as compared with the conserved glutamic acid residues of CphB proteins (8); therefore, it seems on the other hand more likely that Glu 185 is catalytically active.
CphE was active when expressed heterologously in E. coli but was not secreted by recombinant cells. The finding of an N-terminal leader peptide in the amino acid sequence deduced from cphE (Fig. 9) suggested an export mechanism for CphE with specific recognition of the signaling peptide and cleavage FIG. 10. Similarity of the N-terminal leader peptide of CphE (deduced amino acid sequence) from P. anguilliseptica strain BI to N-terminal amino acid sequences of other proteins. Amino acids present in two or in all sequences depicted are shaded in gray. At the amino acid residue positions Ϫ1 and Ϫ3, short chained amino acids are present (potential recognition site for peptide cleavage). A hydrophilic arginine residue and serine residues preceding a hydrophobic leucine-rich domain were found in CphE and in the potential chemotaxis transducer of P. aeruginosa PAO1 (49). In the sequence of the Kex1 protein (precursor of a subtilisin type serine protease from Kluyveromyces lactis (50,51)), two serine residues followed by a leucine and isoleucine-rich hydrophobic domain were detected as well (compare boxed hydrophobic sequences).
FIG. 11. Amino acid sequence similarity of CphE to intracellular CGPases (CphB) and other proteins. The cyanobacterial CGPases form a phylogenetically related group. The strongest similarity of CphE was found to a hypothetical protein of C. crescentus (32% identity over 325 amino acids, NCBI data base search). The isolated position of CphE among CGPases is revealed by the distance between CphE and any other related protein, including the ␣-aspartyl dipeptidase (PepE) from S. typhimurium, previously described as a related protein of intracellular CGPases (8).
of the leader sequence during export in P. anguilliseptica. Similarities of this leader peptide to that of a potential chemotaxis transducer identified in the Pseudomonas aeruginosa PAO1 genome (49), which is according to its function most probably located in the cytoplasmatic membrane, support the assumption that CphE is also membrane-directed. However, the amino acid sequence of the N terminus of CphE produced in E. coli did not deviate from that of native mature (extracellular) CphE, indicating that the suggested leader peptide sequence might be cleaved off in the cytoplasm of E. coli cells.
CGPases are most probably commonly employed enzymes for degradation of a widespread, and therefore, in cases of biomass degradation, often released biopolymer. This is indicated by its high specificity and affinity toward CGP-like material. The enzyme makes the dipeptide building blocks quickly available to cells that possess appropriate proteins for the uptake or further cleavage of ␤-linked amino acid dimers. It should be emphasized that CphE is an extracellular enzyme. It is therefore not involved in the mobilization of intracellular storage polymer CGP; for this, CGP-accumulating bacteria possess intracellular CGPases referred to as CphB (8).