Expression of the Streptomyces Enzyme Endoglycosidase H in Escherichia coZi*

Endoglycosidase H is one of a large number of en- zymes secreted by Streptomyces plicatus and other Streptomyces species. When the structural gene for this enzyme is introduced into Escherichia coli attached to the plasmid pBR-322 or Charon 4 phage, the enzyme is synthesized and is found in the periplasmic space, cul- ture medium, and cells. Attachment of the W - 5 lac promoter to a site in the plasmid adjacent to the Streptomyces insert stimulates enzyme synthesis as much as 100-fold. This result demonstrates that transcription of the Streptomyces gene can be initiated from sequences outside of the Streptomyces insert. Initiation of transcription on a Streptomyces promoter is also a sug- gested but unproven possibility. “signal

Endoglycosidase H, an enzyme secreted by Streptomyces plicatus, has been a major analytical tool in studies of the processing of cell surface glycoproteins. Endo H' hydrolyzes with great efficiency the glycosidic bond between the two Nacetylglucosamine residues of "high mannose" oligosaccharides. Its interesting substrate specificity requires, in addition to the diacetyl chitobiose linkage, additional mannose residues arranged in the specific pattern found in certain eukaryotic glycopeptides ( 1,2).
Endo H was isolated from culture filtrates of S. plicatus by 4). They characterized it as a s m d stable protein (Mr = 27,000). The enzyme is resistant to hydrolysis by Pronase or denaturation by SDS but is inactivated by heat above 60 "C. On sizing columns the protein behaves as a globular monomeric polypeptide. Its amino acid analysis has been published and the NH2-terminal amino acid sequence has recently been determined. ' We became interested in endo H synthesis and secretion for several reasons. A primary aim was to begin a study of enzyme * This work was supported in part by a grant 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. secretion by Streptomyces. These soil organisms secrete an interesting but bewildering array of enzymes which play a major role in the breakdown and reutilization of organic compounds in soil. Very little is known concerning the factors which regulate the synthesis and secretion of these enzymes. For initial studies, it seemed preferable to choose a constitutive enzyme since a Streptomyces enzyme with complex regulatory factors might not be expressed in Escherichia coli. Endo H seemed to be an ideal enzyme since it is a small, stable protein whose synthesis appears to be constitutive (3).
A second aim in cloning endo H into E. coli was to simplify the method for its preparation. Although the purification scheme developed by Tarentino et al. (4) is excellent, it is time-consuming and laborious. High level expression in E. coli and secretion into the periplasmic space should simplify preparation of the enzyme.

EXPERIMENTAL PROCEDURES
Growth of Streptomyces-S. plicatus was obtained from Dr. Frank Maley (Developmental Biochemistry Laboratories, New York State Department of Health, Albany). To maintain cultures, spore suspensions were prepared by gentle scraping in the presence of a small volume of sterile 1% Triton. Spores were sedimented for 15 min at 3000 rpm, resuspended in water, and used to inoculate liquid media or to streak on agar plates. Agar plates for routine passaging contained 0.5% soluble starch, 0.1% NZ-amine, A, 0.1% nutrient broth (Difco), 0.1% yeast extract, and 1.2% agar. To encourage sporulation, plates were seeded sparsely. Since vegetative cultures do not remain viable for extended periods, it is important to induce sporulation in cultures being carried for more than a week. For growth on defined media and for [35S]sulfate labeling of Streptomyces, modifications of the medium of Jeuniaux (5) (J medium) were employed with various carbon and nitrogen sources substituted for chitin. For example, the ["'S]sulfate labeling medium contained (per liter): 0.5 g of NaCI, 0.5 g of NH4CI, 0.7 g of K*HPO4,0.3 g of KH2P04, 0.5 g of MgCh, 0.01 g of FeCL, and 5 g of glucose (added after autoclaving). As noted in the text, the composition of this medium was varied and minimal E. coli media were used from time to time for specific experiments. Minimal E. coli media used in this study were M-9 (0.6% Na2HP04, 0.3% KH*P04, 0.05% NaC1,0.1% NKCI) or OM (1% K2HP04, 0.45% KH*P04, 0.01% MgS04, 0.1% (NH&S04, and 0.05% Na-citrate). The M-9 medium was supplemented after autoclaving with 2 m~ MgS04. Both media were also usually supplemented with proline (20 pg/ml), thymine (20 pg/ml), and thiamine (0.1 pg/ml). The usual carbon source was 0.3% glucose. For [35S]sulfate labeling experiments, chloride salts were used in place of the sulfates. For growth with minimal sulfate supplementation, Na2S04 (10 pg/ml) was added to either sulfate-free M-9 or OM.
Antibody Preparation-Endoglycosidase H was assayed and prepared as described by Tarentino et al. (4). The batch of enzyme used for antibody preparation showed no detectable impurities when analyzed by SDS-acrylamide electrophoresis. NH2-terminal amino acid analysis also indicated that the protein did not contain detectable impurities.* Rabbit antibody to endo H was prepared by subcutaneous injection of 250-pg samples of enzyme with an interval of 3 weeks between injections. The first injection was in complete Freund's adjuvant and the second was in incomplete Freund's. Two weeks after the second injection, the rabbit was bled and y-globulin was prepared by the method of Livingston (6). Another rabbit was used for preparation of immune y-globulin reactive toward one of the chitinase enzymes of S. pZicatus3 (5). The latter y-globulin was used as a control in immune precipitation studies. By titrating enzymatic activity with y-globulin in the presence of Staphylococcus A, it was found that the anti-endo H preparation had a binding capacity of 22 ng of endo H/pg of y-globulin and that endo H and chitinase antibodies did not cross-react.
DNA Extraction and 20-kb Fragment Preparation-DNA was extracted from S. plicatus by a modification of the method used by Shapiro and co-workers for DNA extraction from Caulobacterium (7). A gene library of S. plicatus DNA was constructed using the methods described by Maniatis et al. (8). All restriction enzymes used were from New England Biolabs. S. plicatus DNA was digested in separate reactions with the restriction enzymes Hue I11 and Alu I under conditions, determined empirically for each enzyme, for partial cutting of the DNA. The optimal length of DNA fragments for the construction of a library is 20 kb. The conditions for the Hue 111 (Lot No. 9) were 0.125 uNts/lO pg of DNA digested for 1.5 h at 37 "C and those for A h I (Lot No. 9) were 0.5 unitdl0 pg digested under similar conditions. The DNA was also digested with a 2-fold higher and a 2fold lower concentration of enzyme. The digested DNA was pooled, phenol-extracted, and fractionated on a linear 15-40% sucrose gradient. Individual fractions were analyzed by agarose gel electrophoresis to determine the length of the DNA in each fraction. The fractions containing DNA of 16-22 kb in length were pooled and ethanol-precipitated. The DNA was incubated with Eco RI methylase (New England Biolabs), followed by ligation of Eco RI linkers to the blunt ends (8). The DNA was digested with Eco RI, and the 20-kb DNA was purified on a sucrose gradient.
Preparation of Charon 4 Arms-Charon 4 DNA was prepared by the method of Maniatis et al. (8). Briefly, Charon 4 DNA was digested with Eco RI, incubated at 42 "C for 2 h to allow annealing of the phage arms, and fractionated on a linear sucrose gradient. Appropriate DNA fractions were pooled and concentrated by ethanol precipitation.
Construction of Recombinant Phage-S. plicatus DNA was ligated to Charon 4 DNA at 15 "C with T4 ligase (Lot No. 16). The resulting hybrid DNA was incubated with an in vitro phage packaging reaction (9). E. coli (strain DB 4548) was infected with the recombinant phage and 5 X lo4 plaques were obtained/pg of S. plicatus DNA. When the plaques were tested for S. plicatus DNA by the sensitive method of Woo (10) using nick-translation-labeled S. plicatus 16-to 20-kb DNA as a general probe, approximately half of the visible plaques scored as positive. About 13% of the plaques were lac' but this number decreased to about 1% when the library was expanded according to Maniatis et al. (8). EH-5 was picked from the unexpanded library.
Screening of Phage PZaques-Phage plaques were screened by the method of Broome and Gilbert (11). Plastic discs were coated with anti-endo H (60 pg/ml) and incubated in contact with the phage plaques at 4 "C for 2 h. The discs were then stained with 1251-labeled anti-endo H (11) and exposed to x-ray fh. Duplicate discs were examined. One plaque was picked, purified, and shown to be positive for endo H enzymatic activity by direct testing of a lysate.
Plaques or bacterial colonies were tested directly for endo H activity. Plaques were picked and extracted briefly with endo H substrate (3) in 0.15 M citrate, pH 5.0. Cells taken from colonies were extracted with 5 pl of 0.1 M Tris, 0.1 M EDTA, pH 8, before addition of substrate.
Subcloning of EH-5 DNA-The phage has been prepared on a large scale on plates and in liquid medium. Following phage purification DNA (8) was extracted with phenol, precipitated with ethanol, used for analysis with restriction enzymes, and subcloned into pBR322. Both Barn HI and RI digests of EH-5 were cloned into pBR322 (see Figs. 1 and 2). Bum plasmids were tested for loss of resistance to tetracycline as well as endo H expression. Since ali 'XI" recombinant plasmid colonies were resistant to both ampicillin and tetracycline, colonies were tested directly for endo H (see above). Other plasmids were prepared from the two basic RI and Barn plasmids, pEHR and pEHB, by restriction enzyme treatment followed by religation. Plasmid pEHR3.2 was derived by Pst I digestion of EHR followed by ligation at high dilution. The 1.6-kb fragment containing endo H was recovered from agar (12) and was subcloned into pBR322 which had been digested with Eco RI and Bum. The resulting plasmid is pEHB1.6. Plasmid pEHB1.06 was prepared by Sal I digestion and religation of pEHB1.6 at a high dilution. In each P. Robbins, unpublished data. case plasmids were mapped with restriction enzymes. Endo H levels were measured quantitatively in growing HB 101 cells following transformation (see below). In all cases transformed cells were grown in the presence of ampicillin (100 pg/ml) or tetracycline (20 pg/ml).
The UV-5 lac promoter was prepared from plasmid pKB258, kindly furnished by Dr. M. Ptashne. The 205-base pair promoter was isolated on an acrylamide gel following Eco RI digestion of the plasmid and was ligated to the appropriate plasmid which had been digested with Eco RI. The DNA sequence of the promoter, which reveals the presence of a Pvu I1 restriction site at position 15 of the promoter, was furnished by Dr. W. Gilbert (Biological Laboratories, Harvard University).
Measurement of Tris EDTA-extractible Endo H Levels in Plasmid-carrying E. coli-Overnight cultures were diluted 1:100 in Luria broth (1% tryptone 0.5% yeast extract, pH 7) containing the appropriate antibiotic; either ampicillin (100 pg/ml) or tetracycline (20 pg/ ml). Cultures were shaken at 37 "C and were chilled rapidly in ice water when the A approached 1.0. Cells were sedimented by centrifugation at 12,000 rpm for 5 min and extracted by either a standard EDTA HzO shock or by Tris EDTA. For EDTA H20 shock, cells were washed quickly with half the original volume of 20% sucrose 0.01 M Tris-C1, pH 7.5, and resuspended in 2% of the original volume of sucrose Tris. EDTA treatment and Hz0 extraction were then carried out as described by Koshland and Botstein (13). For Tris EDTA extraction, the initial cell pellet was washed quickly with 0.5 volume of PBS (0.8% NaC1,0.02% KC1, 0.12% Na2HP04, and 0.02% KH2P0,) and was resuspended in 1% of the original volume of 0.1 M Tris, 0.1 M EDTA, pH 8. After incubation for 10 min at 5 "C, cells were removed by centrifugation and the supernatant fluid was used for enzyme assay. The same amount of enzyme is extracted by both methods. Furthermore, SDS-acrylamide gels of the two extracts display similar protein profiles. Table I shows the level of enzyme secreted under a number of different growth conditions. The level of enzyme is approximately the same whether the organism is growing slowly on a poor carbon source, such as glucose, or growing more rapidly in the presence of yeast extract and chitin. In contrast, the chitinase complex of enzymes (5) is synthesized only when the organism is grown on chitin as a carbon source. We have occasionally noticed a slightly higher level of endo H formation when S. plicatus is grown on N-acetylglucosamine (see Table I). In spite of the facts that endo H hydrolyzes a diacetyl chitobiose linkage, the primary linkage in chitin, and that S. plicatus is able to elaborate high levels of chitinase activities, we have confirmed the suggestion of Tarentino   Spores were germinated and cells grown in J medium supplemented where indicated with 0.4% glucose, 0.4% acetylglucosamine, 0.1% yeast extract, and 1% milled crab shell chitin. After growth with shaking at room temperature for 1 week, the mycelium was removed by filtration and solid (NH4)ZSOd was added to saturation. After standing overnight at 4 "C, the precipitate was recovered by centrifugation, dissolved in a minimal amount of 10 mM Tris, pH 7.5, and dialyzed against the same buffer. The thoroughly dialyzed preparation was assayed for endo H (3) and for protein (15).

Expression of Endo H During Lytic Growth of X EH-5-
EH-5, the derivative of Charon 4 with the 12-kb Streptomyces endo H insertion, grows without restriction in several strains of E. coli and expresses low levels of endo H. A partial restriction enzyme map of the Streptomyces insert in EH-5 is shown in Fig. 1, oriented left to right as usually defined for the X genome. Given the fact that transcription for endo H formation in EH-5 is from left to right (see below) and that transcription from the major relevant promoter (PL) is in the opposite direction, it seems possible that endo H transcription in EH-5 infected cells is being initiated from a Streptomyces promoter. Transcription from another X promoter is also a possibility.

Expression of Endo H in Plasmid-carrying E.
coli-Since the assay for endo H is quite sensitive and E. coli has no detectable enzymatic activity of this type itself, it was possible to measure enzyme expression in plasmid-carrying lines accurately. Approximately equal amounts of activity can be released from cells with either a standard shock treatment, as described by Koshland and Botstein (13), or by treatment of cells with 0.1 M Tris, 0.1 M EDTA, pH 8. Enzyme is also found in the growth medium and in the cells. The medium and intracellular activities are currently under investigation and will be the subject of a future publication.
Two basic plasmids were recovered by cloning the 4.3-kb RI and 4.9-kb Bum HI DNA fragments from EH-5 into pBR322 (see Fig. 1). Various derivatives of these two plasmids were prepared as described under "Experimental Procedures." Enzyme levels were measured in growing E. coli HB-101 cells carrying all of these plasmids. Tris EDTA extract enzyme levels and restriction maps of the plasmids are presented in Fig. 2. Maps are shown opened at the Pvu I 1 site of pBR322 and an agarose gel showing plasmid DNA cut with Pvu I 1 is shown in Fig. 3. A more detailed map of the 1.6-kb RI-Bam fragment is presented in Table 11. The presence of two lac promoters in pEHBl.Glac2 oriented as indicated in Fig. 2   shown by acrylamide gel electrophoresis of Puu I1 digestions of this plasmid. As expected, Pvu I1 cuts out the extra 205base pair promoter (data not shown). Enzyme levels were calculated assuming equal specific activities for the M, = 27,000 Streptomyces enzyme and the M , = 30,000 enzyme produced in E. coli. We are purifying the E. coli enzyme to determine both its specific activity and substrate specificity. Transcription may be initiated on the Streptomyces insert and/or on pBR322 sequences. At least one major point is clear. Since insertion of the UV-5 lac promoter at the RI sites of either the EHB1.6 or EHR3.2 plasmids leads to an increase in enzyme synthesis to a constant very high level, it is evident that transcription occurs in the RI to Bum direction on the 1.6-kb fragment. When the UV-5 lac promotor is inserted backwards (pEHB1.6 cal and pEHR3.2 cal), promoting transcription away from the Streptomyces insert, enzyme synthesis is either left near the basal level or is depressed below the basal level. It is interesting that the same high level of enzyme is produced under the influence of the lac promoter whether transcription is taking place in a clockwise (EHBl.6) or a counter-clockwise (EHR3. . The culture (25 ml) was centrifuged at 10,000 rpm for 5 min, washed with sulfate-free M-9, and resuspended in 500 ml of M-9 supplemented as indicated in text. Carrier-free ("'S)sulfate (5 mCi) was added and the culture was shaken at 37 "C for 7 h. Cells were recovered by centrifugation at 10,000 rpm for 5 min, washed with 100 ml of PBS (0.8% NaC1, 0.02% KCI, 0.12% NazHPOr, 0.02% KH2POr), and treated with 6 ml of 0.1 M Tris, 0.1 M EDTA, pH 8. After centrifugation the supernatant fluid was lyophilized, redissolved in H20 to 0.5 ml, and loaded on a Sephadex G-75 column (0.7 x 48 cm). Elution at 5 "C was carried out with PBS diluted with water ( 1 2 ) at a rate of about 2 ml/h. Aggregated and high molecular weight proteins eluted between 6 and 8 ml (fraction 1). Low molecular weight proteins, including endo H (fraction 2), were eluted between 9 and 13 ml, and inorganic sulfate between 17 and 21 ml. one and two lac promoters give rise to about the same level of enzyme in growing cells, significantly higher levels of endo H were observed in stationary cultures carrying the plasmid with two lac promoters (data not shown).
Since coding for the protein probably requires about 800 base pairs, the AUG for initiation of amino acid sequence coding must occur in the first 800 base pairs beyond the RI site on the EHB1.6 plasmid. Since a Sal plasmid carrying only the first 1060 base pairs, EHB1.06, does not promote the formation of complete enzyme, it seems likely that the initiating AUG lies between positions 200 and 800. Since the six NH2-terminal amino acids are now known, it should be possible to locate the coding region and identify the endogenous Streptomyces promoter by direct DNA sequencing.
Size of the Enzyme Extracted from E. coli-Maley and coworkers have purified S. plicatus endo H to homogeneity and have shown that it behaves as a monomeric polypeptide with a molecular weight of 27,000 (3). When the enzyme is extracted from plasmid-carrying lines of E. coli and sized on a Sephadex G-75 column, it behaves as a monomeric polypeptide of approximately the same size (data not shown).
When the E. coli enzyme is labeled with [?S]S042or [:%I methionine, precipitated with anti-endo H antibody, and examined by SDS-acrylamide gel electrophoresis, however, it is clearly seen to be 2,000-3,000 daltons larger. We have labeled the protein under a variety of conditions and attempted to induce cleavage to the M , = 27,000 form without success. In the experiments shown in Fig. 4, the enzyme was labeled with [:"S]sulfate and extracted from the periplasmic space with Tris EDTA. Aliquots were incubated alone or with S. plicatus cells or culture medium for 15 h a t 24 "C. Half of each sample was immunoprecipitated with anti-endo H or control y-globulin.
Although no E. coli protein reactive with our anti-endo H serum smaller than 29,000-30,000 has been observed, larger polypeptides have occasionally been seen. Precipitation of labeled Streptomyces culture medium with our anti-endo H reagent also reveals larger polypeptides as well as the expected M , = 27,000 enzyme. At the present time, we are attempting to determine which of these polypeptides, if any, are related to endo H.

DISCUSSION
The constitutive nature of endo H synthesis in Streptomyces as well as its extreme stability in the face of proteolytic enzymes and denaturing agents have probably contributed to our detection of the enzyme following cloning of Streptomyces genomic DNA in E. coli. The gene for endo H, after cloning in E. coli, may be transcribed from a promoter in the Streptomyces DNA or from promoters in the vector DNA. In the Charon 4 clone, EH-5, transcription of the endo H gene must have taken place in the direction opposite from transcription initiated at PI,, the major relevant h promoter. This observation suggests that transcription may occur from a Streptomyces promoter, although phage transcription obviously cannot be ruled out. The level of endo H expression, in X EH-5infected cells is relatively low, in the range of 10-100 molecules/infected cell (data not shown).
In the plasmid subclones, it seems probable that pBR322 promoters play a role in endo H transcription. Of the two Streptomyces plasmids containing complete pBR322 genomes, the plasmid with the higher level of expression (EHR) has its endo H sequence placed in such a way that transcription continuing from the p-lactamase promoter at the plasmid Hind111 site would immediately give a correct transcript of the enzyme. In EHB, the endo H gene is oriented in the opposite direction with respect to pBR322. In this case tran-scription of the endo H gene continuing from the plasmid promoter for the tetracycline resistance (tet) gene (located at the Hind111 site) may increase gene expression as evidenced by the %fold lower level of expression seen when this promoter is deleted (EHB1.6). The obvious factors which could be of importance in the level of endo H expression from these plasmids (EHR, EHB, and EHB1.6) are the different levels of promotion from the /3-lactamase and tet promoters and the distance between the promoter and the endo H gene. It w i l l be interesting to determine enzyme production when EHB and EHR fragments are each reversed in the plasmid.
In pEHB1.6 the tet plasmid promoter has been deleted and the /3-lactamase promoter is oriented away from the endo H gene. In spite of this, the endo H gene is expressed. This suggests promotion from a Streptomyces promoter, but, as in the phage clone, low level transcription from some yet unidentified pBR322 promoter cannot be ruled out. Another element not taken into account here is plasmid number per cell which undoubtedly varies from lime to Iine. The powerful promotion which results from insertion of the UV-5 lac promoter is similar to the stimulation seen when this element is inserted in front of the h repressor gene (14). Although UV-5 contains a ribosome binding site it is clear that the ribosome binding site is probably specified by the Streptomyces genome since translation is initiated 200-800 base pairs from the point of lac insertion, and the protein made is the same size whether or not the lac promoter is present.
It is certainly not yet clear whether the larger polypeptide formed in E. coli is the result of lack of cleavage of the Streptomyces "leader" sequence (the sequence associated with secretion) or is the result of lack of cleavage by other secreted Streptomyces enzymes. To date we have not been able to bring about cleavage of the E. coli enzyme by Streptomyces cells or medium but this may simply be the result of inadequate conditions. We have recently become aware that the pattern of Streptomyces secreted proteins precipitated by our antibodies are variable and sensitive to culture conditions. We plan to examine these proteins before drawing conclusions concerning the difference between the E. coli and Streptomyces enzymes.