Permeability Properties of Chemically Modified Porin Trimers from

The pore-forming protein of the outer membrane of Escherichia co& porin, was chemically modified with acetic anhydride, succinic anhydride, and glycinamide. Extensive modification of amino groups of the func- tional pox-in trimers caused reduced diffusion rates of the negatively charged solutes such as p-nitrophenyl phosphate and AMP, but did not reduce significantly the diffusion of positively charged molecules carbo- benzoxy-glycyl-prolyl-a&nine-p-nitranilide and tos-yl-glycyl-prolyl-a&nine-p-nitranilide. Modification of carboxyl groups of trimers caused decreased diffusion rates of the positively charged solutes more signifi- cantly than the diffusion rates of negatively charged solutes. The results suggest that the ionic interactions play an important role for the diffusion of charged solutes through the porin pore. The diffusion ofp-nitro- phenyl a-D-glucoside, an uncharged solute, was not influenced significantly by modification of either amino or carboxyl groups. This observation suggests that modifications only occurred in areas

The pore-forming protein of the outer membrane of Escherichia co& porin, was chemically modified with acetic anhydride, succinic anhydride, and glycinamide. Extensive modification of amino groups of the functional pox-in trimers caused reduced diffusion rates of the negatively charged solutes such as p-nitrophenyl phosphate and AMP, but did not reduce significantly the diffusion of positively charged molecules carbobenzoxy-glycyl-prolyl-a&nine-p-nitranilide and tosyl-glycyl-prolyl-a&nine-p-nitranilide.
Modification of carboxyl groups of trimers caused decreased diffusion rates of the positively charged solutes more significantly than the diffusion rates of negatively charged solutes. The results suggest that the ionic interactions play an important role for the diffusion of charged solutes through the porin pore. The diffusion ofp-nitrophenyl a-D-glucoside, an uncharged solute, was not influenced significantly by modification of either amino or carboxyl groups. This observation suggests that modifications only occurred in areas outside of the narrowest portion of the pore or, alternatively, that amino and carboxyl groups are exclusively located at noncylindrical area of the pore. The structural integrity of the acetylated and the succinylated trimers seemed well preserved.
On the other hand, modification of carboxyl groups decreased the thermal stability of trimers and extensive modifications caused the dissociation of trimers into monomers at 37 "C. and other solutes of low molecular weight but not the molecules of hydrophobic nature such as bile acids, detergents, or dyes, we proposed that the pores might be fiied with water and the interior of the pores must be covered with hydrophilic amino acid residues (5,9). Thus, we are interested in studying the possible effects of chemical modifications of amino and carboxyl groups of the functional porin trimers on the permeability properties of the pores, and on the structural integrity of the trimers.
In this paper, we describe acetylation and succinylation of amino groups and amidation of carboxyl groups of porin trimers, and the effects of such modifications on the permeability properties of the pores in the reconstituted vesicle membranes.

Bacterial
Strain and Culture Conditions-The bacterial strain used throughout this study was E. coli B, producing a single species of porin (6, IO). Cells were grown in a medium containing 1% Bactotryptone, 0.5% yeast-extract, and 0.5% NaCl in either Erlenmeyer flask or in a lOO-liter jar fermentor under vigorous aeration. Preparation of Porin-Porin trimers were prepared according to the procedures described earlier (7) except that 10 mu phosphate buffer, pH 7.0, was used throughout instead of 10 mu Tris-HCl buffer, pH 7.5.
Acetylation of Porin Trimer-The purified trimers (59.5 nmol) in l-ml solution containing 45 mu sodium phosphate buffer, pH 6.0 or 7.0, 0.1% SDS,' and 3 mM NaN3 were mixed with a total of 56.5 pmol of [Wlacetic anhydride (specific activity 0.169 mCi/mmol) by adding one-sixth volume of the reagent at every 10 min up to 60 min, and were kept at 25 "C for 60 min. The pH of the mixture was maintained to either 6.0 or 7.0. The protein in the mixture was separated from unreacted free reagents by gel filtration on a Sephadex G-25 column (0.9 x 56 cm, Pharmacia, Fine Chemicals), equilibrated with a solution containing 0.1% SDS and 10 mM sodium phosphate buffer, pH 6.0 or 7.0, and the column was eluted with the same solution. Cbz-Gly-Pro-Argnitranilide, carboxylbenzoyl-glycyl-prolyl-arginine-p-nitran~ide; tosyl-Pro-Pro-Arg-nitranilide, tolylsulfonyl-glycyl-prolyl-arginine-p-nitranilide.

8025
at 80,000 X g for 30 min in order to replace SDS with Triton X-100. This material (6.4 mg of protein) was mixed with 2 M glycinamide (pH was maintained a t about 5 with NaOH) a t a final molar ratio of protein to glycinamide of 1 to 28,600. The mixture was divided into six equal portions and the pH of each portion was adjusted to 4.75, 5.0, 5.3, 5.6, 6.0, or 6.8, with either 1 N HCI or NaOH. The amidation reaction was started by adding dropwise, 100 pl of freshly prepared EDAC (4.9 X mol) under controlled pH. The reaction was stopped after 2 h a t 25 OC by adding 1.5 volumes of 1 M acetate buffer of appropriate pH. The mixture was dialyzed, 10 min later, against a large excess of distilled water a t 4 OC for 36 h and the contents of the dialysis bag were collected by centrifugation a t 80,000 X g for 60 min, suspended in the solution containing 1% SDS, 0.4 M NaCI, 5 mM EDTA, 0.0510 mercaptoethanol, 10 m~ sodium phosphate buffer, pH 7.5, and 3 m~ NaN3. The centrifuged supernatant (80,000 X g for 60 min) contained the modified porin trimers (7). The number of carboxyl groups modified was calculated from the increment of glycine residues as determined by amino acid analysis after hydrolysis.
Thermal Stability of Porin Trimer-Porin trimers (50 pg) in 50 pI of 8 mM sodium phosphate buffer, pH 8.0, containing 2% SDS were placed in an Eppendorf Microtube (capacity 1.5 ml) (Eppendorf Geratebau), heated in a water bath a t desired temperature for 5 min, and chilled immediately in crushed ice. A portion of the sample (4 pg of protein) was subjected to acrylamide gel electrophoresis in SDS

A B C D E
FIG. 1. Electrophoretic mobilities of the modified porin trimers. The buffer system for electrophoresis in SDS was described (18). The total acrylamide concentration was 10%. A, native porin trimers (3 pg); B, porin monomers (3 pg), prepared by heating the trimers at 100 "C for 5 min in "sample buffer" (18); C, acetylated porin trimers (2 pg). The trimers were acetylated at pH 7.0 so that 27 amino groups/trimer were found to be modified. D, succinylated porin trimers (3 pg). The trimers were succinylated at pH 8.0 as described above producing 41 modified amino groups/trimer. E, amidated porin trimers (3.5 pg). Carboxyl groups of trimers were modified a t pH 5.6 as above yielding 60 modified residues/trimer. The figure shows only relevant portions of the stained gels. and the stained protein bands were traced by Shimadzu TLC-Scanner LS-900 (Shimadzu Seisakusho Ltd.).
Permeability Assay-Reconstitution of vesicle membranes and the rate assay of solute diffusion were described (12).
Preparation of Anti-porin Antibody a n d Immunodiffusion Test-Liposomes containing purified porin trimers were made as reported earlier (13). from 30 pmol of phospholipids extracted from E. coli B and 3 mg of purified porin trimers, except that lipopolysaccharide was omitted and 0.1 M NaCl was used as suspending medium.
Rabbits (Albino J. W. female, about 2.5 kg) were injected subcutaneously with liposomes containing 500 pg of porin, and 10 days later, they were boosted with liposomes containing 1 mg of porin. Twenty days after the first immunization, the rabbits were bled and serum was obtained. Serum was fractionated with 45% saturation of (NH4)2S04 and the precipitates were dissolved in a small amount of 0.1 M NaCl containing 10 mM sodium phosphate buffer, pH 8.0. The precipitates were dilayzed against a large excess of the same buffer. A portion of the dialyzed material (410 mg of protein) was subjected to gel filtration by a Sepharose 4B column (2.5 X 90 cm) equilibrated with the above buffer and the column was eluted with the same buffer. The eluates corresponding to IgG fraction were pooled, concentrated by dialysis against dry Ficoll 400, and dialyzed against a large excess of 50 mM sodium phosphate buffer, pH 8.0, containing 0.15 M NaCl a t 4 "C overnight.
Double diffusion immunoprecipitation test was carried out in a agar plate (1.0 mm thickness) containing 0.15 M NaCI, 0.5% SDS. and 3 mM NaN:, at room temperature for 24 h.
Other Methods-Circular dichroism spectra were recorded with a JASCO 5-20 automatic recording spectropolarimeter as reported (7). The mean residue molecular weight of porin was calculated from the amino acid composition (9). Amino acids were analyzed as reported earlier (9) except that 6 N HCI instead of 4 N methanesulfonic acid was used for hydrolysis. Number of 0-acetyltyrosine residue of wetylated porin was quantified according to the method described by Simpson et al. (14). Phospholipids were extracted from E. coli B by the method of Bligh and Dyer (15), and were quantified by the method of Marinetti (16). Protein was quantified either by the method of Lowry et al. (17), or by absorption a t 278 nm. Absorption coefficient (A?;:",,,) of native, acetylated, succinylated, and amidated porin trimers were 1.69, 1.69, 1.67, and 1.84, respectively. Procedure for acrylamide gel electrophoresis in SDS was described earlier (18).

RESULTS
Modification of Amino Group of Porin Trimer-We have used acetic anhydride and succinic anhydride for modification of amino groups because of the following reasons: (i) since the reagents are relatively small, we hoped that they can modify the amino groups located deep within the pores; (ii) acetic anhydride converts positively charged amino groups into electrically neutral groups and should confer hydrophobicity to the pore wall. (iii) Succinic anhydride replaces positively charged amino groups with negatively charged carboxyl groups. When porin trimers were treated with acetic anhydride at pH 6.0 or 7.0, 15 or 27 amino groups/trimer, respectively, were acetylated. Similarly 40 residues were modified by succinic anhydride at pH 8.0. It is unlikely that these reagents modified the amino groups embedded in a lipophilic environment, since less than one amino group of the trimer (total 57 residues) incorporated ['4C]phenylisothiocyanate, a lipophilic reagent that only attacks unprotonated amino groups, as examined using isolated outer membrane at pH 7.0, at 37 "C for 2 h.2 Porin trimers treated with acetic anhydride or succinic anhydride a t appropriate conditions appeared as single protein bands migrating slightly above (relative mobil-M. Tokunaga and T. Nakae, unpublished result. TABLE I Diffusion of negatively charged and neutral solutes through the vesicle membranes reconstituted from modifiedporin trimers choline and 5 nmol of porins as reported earlier (12). Membrane procedure were similar to that described in earlier publication (see vesicles for assaying the diffusion rate of p-nitrophenyl phosphate, legend to Table 11  ' The total number of amino and carboxyl groups per trimer was 57 and 147, respectively. e The fluctuation of relative diffusion rate (56) of NADPH was probably due to lower permeability efficiency values for this solute.
The permeability efficiency (PA) values presented in the table were the average fS.D. of at least three independent assays.

TABLE I1
Diffusion ofpositively charged solutes through the vesicle membranes reconstituted from modified porin trimers Membrane vesicles were prepared from 6 pmol of egg yolk phosphatidylcholine and 5 nmol of porins. Vesicles were suspended in 200 p1 of a solution containing 50 m~ Tris-HC1, pH 8.0,50 mM NaC1, and 2 mg of trypsin and were passed through a Sepharose CL-GB column as reported earlier (12). Solute diffusion was monitored at 405 nm at the substrate concentration of 5 X M at oH 8.0 at 25 "C.

Porins"
Native porin trimer Acetylated porin trimer at pH 7.0 Succinylated porin trimer Amidated porin trimer at pH 5. ities to dye front were 0.23 and 0.22 for acetylated and succinylated porin, respectively) the native trimers (relative mobility was 0.24) as examined by polyacrylamide slab-gel electrophoresis in SDS (Fig. 1). The result suggested that acetylated and succinylated porins still retained the trimer structure. When the presence of 0-acetyltyrosine residues in the acetylated trimers was determined by the method of Simpson et al. (14) by recording differential spectra before and after NHzOH treatment, we found no detectable 0-acetyltyrosine residue in the modified trimers.
If the reagents modify the rim and/or the interior of the pores, it is conceivable that the vesicle membranes containing the modified trimers have altered permeability properties for certain solutes. The diffusion rate ofp-nitrophenyl phosphate and AMP through the vesicle membranes reconstituted from acetylated porin trimers (15 modified residues), fell to about 62%, and 39%, respectively, of the diffusion rates of these compounds through the membrane containing the native trimers. As the extent of acetylation reached to 27 residues out of 57 available amino groups, the diffusion rates of p-nitrophenyl phosphate and AMP through the membranes containing these porins dropped to 50% and 22%, respectively, of the control ( Table I). The succinylation reaction yielded the trimers, of which 41 amino groups were modified, that allowed the diffusion of p-nitrophenyl phosphate and AMP at only 24% and 17%, respectively, of the rate obtained with native trimers. Conversely, the diffusion rates of positively charged compounds such as Cbz-Gly-Pro-Arg-nitranilide and tosyl-Gly-Pro-Arg-nitranilide through the membrane containing acetylated or succinylated trimers appeared to be 117 and 179% or about 60 and 135%, respectively, of the rates through the native trimers (Table 11). Cause for the different degree of effect of chemical modifications on the diffusion rate of the two positively charged solutes was unexplained. The diffusion rates of p-nitrophenyl-a-glucoside through the vesicles membranes containing the acetylated and the succinylated trimers appeared to be 93 and 86%, respectively, of that through the native trimers (Table I). From the results shown on Table I and 11, the following lines of conclusion may be drawn; (i) the modification of amino groups diminished net positive charge of the rim or wall of the pore, and this resulted in the reduction of diffusion rates for negatively charged solutes. (ii) These modifications had little influence on the diffusion rates of uncharged molecules suggesting that reagents only attacked amino or carboxyl groups located at nonsyrindrical portion of the pores. (iii) The modification frequently enhanced the diffusion of positively charged solutes. Modification of Carboxyl Group of Porin Trimer-The method employed first activates carboxyl groups with EDAC with the subsequent attack on the activated carboxyl group with a nucleophile, glycinamide. Since the treatment of purified trimers with EDAC and glycinamide in the presence of 0.1% SDS or 0.1% Triton X-100 at pH 4.75 resulted in an extensive dissociation of trimers into monomers and since the same treatment in 0.1% Triton X-100 at pH 5.6 produced aggregates of the modified trimers, we used the peptidoglycanassociated porin trimers, rather than free trimers, for modification in the presence of 0.1% Triton X-100. (The trimers are stable under these conditions down to pH 2 at 25 0C).3 When the reaction was carried out at pH 4.75,5.0, 5.3,5.6,6.0, or 6.8, 78, 75, 69, 60, 54, or 27 carboxyl groups, respectively, per trimer were found to be modified. Extensive modification of carboxyl group resulted in the dissociation of a significant fraction of trimers into monomers (Fig. 2), as determined by slab gel electrophoresis in SDS. Porin trimers, modified in up to 60 carboxyl groups, appeared as a sharp single protein band moving slightly more slowly (relative mobility to dye front was 0.18) than native trimers (Fig. 1). The modification reaction carried out without EDAC yielded no detectable reaction product, and the reaction without glycinamide resulted in extensive cross-linking of the trimers and the formation of amorphous aggregates.
Vesicle membranes containing the modified trimers (27 carboxyl groups were modified per trimer) allowed the diffusion of p-nitrophenyl phosphate, AMP, and p-nitrophenyl aglucoside as rapidly as native trimers did (Table I). The vesicles containing the trimers, in each of which 60 carboxyl groups on an average were modified, showed only 31,28, and 13% decrease in their permeability to p-nitrophenyl phosphate, AMP, and p-nitrophenyl a-glucoside, respectively, in comparison with those containing of the native trimers. On the other hand, the diffusion rate of NADPH, a slowly diffusing substrate, through the membrane containing the modified porins increased about 2-to 2.5-fold. The diffusion rates of Cbz-and tosyl-Gly-Pro-Arg-nitranilide through the membrane containing amidated trimers appeared to be 69 and 47% of those in vesicles containing native trimers (Table 11). These results suggested several possible interpretations (see "Discussion"). Porin trimers modified at pH 4.75 conferred little permeability to the vesicles. This result is explained by the fact that trimers are dissociated almost completely into mon-T. Nakae, unpublished result. in terms of molar ellipticity. Porins were dissolved in 50 m~ phosphate buffer, pH 8.0, containing 0.1% SDS, except that amidated porins were dissolved in a solution containing 1% SDS, 0.4 M NaC1, 5 mM EDTA, 0.05% mercaptoethanol, 10 m~ sodium phosphate buffer, pH 7.5, and 3 m~ NaN3. A, native trimers; B, succinylated trimers; C, amidated trimers; D, acetylated trimers. Porins were modified under the conditions similar to that of Fig. 1, and the extents of modifications were determined as identical with the porins used for the experiment in Fig. 1.

T o m p r r o t u r r ( C )
FIG. 4. Thermal stability of modified porin trimers. Native and chemically modified trimers (same samples used for the experiment in Fig. 1) were treated as described in the text. M , the native trimers; A-A, the acetylated trimers; M , amidated trimers. omers (Table I and Fig. 2).
Structural Integrity of the Chemically Modified Porin Trimer- Fig. 3 depicts the CD spectra of acetylated, succinylated, and amidated porin trimers as well as that of native trimers. The CD spectra of the modified porins showed profiies fairly close to those of native trimers, suggesting P-rich conformations (Fig. 3). The molar ellipticity at 217 nm of the acetylated, succinylated, amidated, and the native trimers were calculated to be -7,880, -7,600, -8,350, and -8,180 deg cm2 dmol", respectively.
Since we often observed that the extensive modification of carboxyl groups tend to disaggregate trimers into monomers in a solution containing 1% SDS, 0.4 M NaC1, 10 mM sodium phosphate buffer, pH 7.0, we wondered if the structural integrity of the modified trimers was preserved, although CD spectra showed little difference. When the amidated trimers (60 carboxyl groups per trimer were modified) were heated at various temperatures in the presence of SDS, they began to dissociate at 40 "C, and a temperature causing 50% dissociation under our conditions appeared to be about 48 "C. In contrast, the native, acetylated, and succinylated trimers began to dissociate a t about 62 "C and the complete dissociation was attained a t 70 "C. The temperature causing 50% dissociation of the native and amino group-modified porins were indistinguishable from each other, about 67 "C (Fig. 4). These results clearly indicated that the elevated thermal sensitivity of the modified trimers can be attributed to the introduction of glycinamide onto the carboxyl groups of trimer.
Immunodiffusion Test-Antibody raised against native POrin trimers was used for agar gel precipitation with modified and native porins. As shown on Fig. 5, anti-trimer antibody formed precipitation lines with the native and the modified trimers but not with the heat-dissociated monomers. Although immunoprecipitation lines of the acetylated and succinylated trimers fused completely, the lines between the native trimers and either the acetylated or succinylated trimers produced a fused line and a spur, suggesting that modification of amino groups blocks at least one immunodeterminant present in the surface of the native trimers (Fig. 5 ) . Similarly, the precipitation lines between native and the amidated trimers produced one fused line and a spur, and that between carboxyl and amino group-modified trimers produced one fused line and one spur over each well (data not shown). The results suggested that the modifications altered a part of trimer surface, but the modified trimer unlike the monomer, still retained a gross conformation close to native trimer.

DISCUSSION
The permeability of small hydrophilic molecules through the outer membrane of Gram-negative bacteria is a passive diffusion event through nonspecific diffusion pores made of porins (5, 6, 13). The amino acid analysis of porins from E. coli (10) and Salmonella (9) showed that the proteins are rich in polar amino acids in spite of the fact that they are intrinsic membrane proteins. This apparent inconsistency is probably due to the fact that porins need relatively large numbers of 'ion of E . coli Porin polar amino acid residues in order to construct the water-filled pores (9). Recently, it was shown that the smallest functional unit of porin is a trimeric aggregate of homologous porin subunits (7, 8,19,20). It was therefore of interest to modify the amino and carboxyl groups of the functional porin trimers and to see the effect of modifications on permeability properties of the pores.
As regards the effects of chemical modifications on the permeability of solutes, we found essentially that the extensive modification of amino groups caused lowered diffusion rates of the negatively charged compounds, and the modification of carboxyl groups resulted in reduced diffusion of positively charged solutes. The modification of amino or carboxyl group had little influence on the diffusion of uncharged molecule. We favor the following interpretation of these results. Since acetylation and succinylation of the trimer makes the net charge of the pore interior and/or the rim more negative, the decreased diffusion rate of the negatively charged solutes are most likely due to electrostatic repulsion between the solutes and the pore wall or rim. The reduced diffusion rates of positively charged compounds through the pore of amidated trimers offered an additional evidence to support the role of electrostatic forces for diffusion of charged molecules. That the reduced diffusion rates of charged solutes were solely due to the physical blockades formed at the interior or the rim of the pores appear unlikely, since the diffusion rates of p-nitrophenyl a-glucoside, a relatively large and uncharged molecule were less altered. However, the result do not exclude a possibility that the modifications only occurred at the rim but not the interior of the pore wall, or that the narrowest portion of pore wall contained little amino or carboxyl groups. The observation that the amidated trimers allowed the less altered or even faster diffusion ofp-nitrophenyl phosphate, AMP, and NADPH and that the acetylated and succinylated trimers allowed accelerated diffusion of positively charged compounds may be explained similarly by the ionic interactions between pore wall or rim and the solutes. The possibility that the decreased diffusion rate of certain solutes through the membranes containing modified porins are simply due to an inefficient incorporation of modified porin trimers into liposome membranes is unlikely, since the diffusion of p-nitrophenyl a-glucoside through these membranes was not reduced.
Modification of carboxyl groups tended to destabilize the integrity of the trimers as shown in Figs. 2 and 4. Several explanations are possible. (i) Negative charges of carboxyl groups form cross-bridges with cationic groups of the neighboring subunit via ionic interactions. (ii) Carboxyl groups are clustered at the periphery of the trimer and they shield the interior from the attack of an ionic surfactants such as SDS.
(iii) Neutralization of the negatively charged groups by amidation strengthen electrostatic repulsions between positively charged groups causing distortion of the trimer structure. (iv) Introduction of glycinamide exerts mechanical distortions on the quarternary structure of the trimers. Possibility i appears unlikely, since the trimers are quite stable in a solution containing SDS and moderate concentration of NaCl (7). We are not happy with explanation ii, since the treatment of the trimers with a cationic surfactant, cetyltrimethylammonium bromide, did not cause a significant dissociation of the trimers, and the trimers modified with taurine, 2-aminoethanesulfonic acid, appeared to be unstable in SDS.* We have no concrete evidence to support or to reject explanations iii and iv. Since the trimers dissociate into monomers along with the increased amidation of carboxyl groups (Fig. 2) but not by the extensive modification of amino groups, it seems reasonable to assume that certain numbers of carboxyl groups play a crucial role in maintaining the quaternary structure of the trimer.