Artificial hybrid protein containing a toxic protein fragment and a cell membrane receptor-binding moiety in a disulfide conjugate. I. Synthesis of diphtheria toxin fragment A-S-S-human placental lactogen with methyl-5-bromovalerimidate.

In order to study the mechanism of entry of plant seed and bacterial toxins into mammalian cells, methods have been developed to synthesize artificial protein hybrid conjugates containing a moiety which binds to a cell membrane receptor and an active fragment of a toxin protein. Utilizing methyl-5-bromovalerimidate, a disulfide cross-linked conjugate of human placental lactogen (hPL) and diphtheria toxin fragment A (toxin A) was synthesized. The reagent was prepared from 5-bromovaleryl nitrile by Pinner synthesis and then used to amidinate hPL. The bromo group thus introduced was converted to S-sulfonate by nucleophilic displacement with 1 M aqueous sodium thiosulfate at room temperature overnight. The S-sulfonated hPL reacted readily with the-SH gorup of reduced toxin A to form a 1 mol/mol of disulfide conjugate in high yield. Thus when reduced toxin A was incubated with a 4-fold excess of the hPL S-sulfonate at 4 degrees and pH 6.5 for 120 h, a conjugate yield of 50% relative to the toxin A input was obtained. Homopolymer formation was negligible and the product was purified by gel filtration on Sephadex G-150. Purity of the conjugate estimated by quantitative analysis of sodium dodecyl sulfate gels was 90%. The toxin A-hPL conjugate retained the activities of both toxin A and hPL, as reported in the accompanying paper. This method of preparing protein hybrid conjugates appeared to have advantages over previous methods utilizing bifunctional reagents with respect to both yield and freedom from homopolymer formation.

In order to study the mechanism of entry of plant seed and bacterial toxins into mammalian cells, methods have been developed to synthesize artificial protein hybrid conjugates containing a moiety which binds to a cell membrane receptor and an active fragment of a toxin protein. Utilizing methyl&bromovalerimidate, a disulfide cross-linked conjugate of human placental lactogen (hPL) and diphtheria toxin fragment A (toxin A) was synthesized. The reagent was prepared from 5-bromovaleryl nitrile by Pinner synthesis and then used to amidinate hPL. The bromo group thus introduced was converted to S-sulfonate by nucleophilic displacement with 1 M aqueous sodium thiosulfate at room temperature overnight. The S-sulfonated hPL reacted readily with the -SH group of reduced toxin A to form a 1 mol/ mol of disulfide conjugate in high yield. Thus when reduced toxin A was incubated with a 4-fold excess of the hPL Ssulfonate at 4" and pH 6.5 for 120 h, a conjugate yield of 50% relative to the toxin A input was obtained. Homopolymer formation was negligible and the product was purified by gel filtration on Sephadex G-150. Purity of the conjugate estimated by quantitative analysis of sodium dodecyl sulfate gels was 90%. The toxin A-hPL conjugate retained the activities of both toxin A and hPL, as reported in the accompanying paper. This method of preparing protein hybrid conjugates appeared to have advantages over previous methods utilizing bifunctional reagents with respect to both yield and freedom from homopolymer formation.
This paper is the first in a series which explores the biochemical and biological properties of artificially constructed hybrid protein-protein conjugates cross-linked by a disulfide bond. The present paper gives details about a general method for the high yield synthesis and purification of such hybrid conjugates. Each of the hybrids which we have made contains the enzymatically active fragment of a seed or bacterial toxin, such as diphtheria toxin A chain, which is capable of inhibiting protein synthesis in cell-free systems but is inactive toward intact cells. The other half of the hybrid is a protein, usually of nontoxin origin, having high affinity for a cell surface membrane receptor.
The mechanism of action of the toxins used as starting materials in these studies has been reviewed in detail (l-3). Each toxin consists of two polypeptide chains cross-linked by a disulfide bond. The A chain or active chain catalytically inhibits eukaryotic protein synthesis inside the cell. The B chain binds the toxin molecules to a specific cell surface receptor. A or B chains alone are nontoxic for cells; however, their separate activities can be measured. Purified A chains can catalytically inhibit protein synthesis in cell-free systems. Purified B chains can bind to their specific cell surface receptors (4). The toxic effect toward intact cells requires combination of the two chains. Recent studies with antitoxin antibodies and with molecules that can compete with the B chain for binding the cell surface receptor have provided evidence indicating that the action of the toxins involves at least three sequential steps: (a) binding of toxin through B chain, (b) entry of the A chain or the whole toxin molecule, and (c) inhibition of protein synthesis by the A chain (l-3). The mechanism by which the A' chain or the whole toxin enter the cell is as yet unknown. At least three toxins, nicked diphtheria toxin, abrin, and ricin fit the above generalities.
Abrin and ricin are similar toxins in that both their B chain and A chain specificities are closely related. The B chains of these toxins bind to galactose-containing residues. The binding specificity for diphtheria toxin has not been reported. The A chains of ricin and abrin inactivate ribosomes apparently by inhibiting the GTPase site on the 60 S ribosomal subunit (5). Diphtheria toxin A chain inactivates elongation factor II by catalyzing the transfer of ADP-ribose moiety from NAD+ to the soluble enzyme (6).
The hybrid conjugates which we have constructed and report on in these papers may be regarded as structural analogues of toxin molecules in which the original binding chain has been substituted by another protein displaying different binding specificities.
Our reasons for studying hybrid protein conjugates containing the active chain of these toxins are as follows. First, there is very little information available about the mechanism of entry of protein molecules which exhibit an obligatory first step of binding to specific cell surface receptors. We will refer to this process as receptor-mediated entry of proteins. We wish to know whether all surface membrane receptors participate 1506 Synthesis of Diphtheria Toxin A-S-S-Human Placental Lactogen in this phenomenon, or whether receptor-mediated entry is limited to a class of receptors displaying unique properties. Second, there is the question of whether the entry process is only dependent on the properties of the B chain or whether A chain properties are also involved. If entry requires only the B chain, then hybrid molecules with competent B chains will direct entry of any hybrid A chain. Such hybrids would have vast biologic potential. Since various cell types display different surface membrane receptors, it should be possible, by choosing the right B chain, to construct hybrid conjugates which are cell type-specific. Such hybrids would constitute an entirely new class of pharmacologic reagents. An obvious application would be cell-type-specific cancer chemotherapy. In the following paragraphs we discuss the problems in synthesizing artificial protein hybrid conjugates and the general methods which we have found to be successful.
Artificial protein hybrid conjugates can be produced by simply adding a cross-linking reagent to a mixture of the protein species. Only small amounts of conjugated hybrid are produced in this manner, most of the material consisting of homopolymers of varying sizes (7). For our purposes we believed it was necessary to make hybrid conjugates in such yields that the species could be purified, unequivocally identified by physical and biochemical methods, and the contamination with homopolymers reliably estimated. This required synthesis in the milligram range and suppression of homopolymer formation.
Homopolymer formation is not a problem fide-linked, like the original toxins. Since one of our species, the A chain of diphtheria toxin, contained a single -SH group, which could be converted to -SSO:,-if desired, it became only necessary to introduce a single --SH or -SSO,group into our putative B chain. When putative B chains contained internal disulfide bonds these were left intact in order to minimize disturbance of the configuration of the binding protein which can lead to loss of binding activity. In part this was based on previous work with insulin (11) and prolactin' ) Our attempts to introduce -SH or -SSO,-groups into proteins using commercially available imidate reagents, methyl-4-mercaptobutyrimidate and dimethyl dithiobis(3,3'propionimidate), met with failure. We found it was very difficult to prevent the attack on the disulfide bonds of protein by the -SH of the mercaptoimidate under the condition of amidination. And it was equally difficult to make S-sulfoimidates from these reagents prior to amidination of pr0teins.l An attempt was then made to synthesize an imidoester Ssulfonate from a haloalkyl nitrile by first converting the halide into S-sulfonate by Bunte synthesis (12) followed by converting the nitrile into imidoester by Pinner's method (13). This also failed due to a solubility problem encountered in the second step. However, by reversing these two procedures we have successfully developed a method to introduce an extrinsic S-sulfonate group into proteins according to the following reactions.
NH,+. Cl-II Br-(CH,),CN + CH,OH + HCl + Br-(CH,),C-OCH, NH,+ NH,+ II II Protein-NH-C(CH,),-Br + S,O,"-+ Protein-NH-C(CH,),-SSO,-+ Br- when naturally occurring proteins with multiple polypeptide chains are reassembled from their subunits in the test tube (8). This is because the heterospecies are held together and preferentially oriented by strong noncovalent forces (8) which direct covalent cross-linking (9). The natural polymeric form is thus kinetically and thermodynamically the preferred species. In contrast, our early attempts to cross-link two polypeptide species which are not naturally associated showed that the preferred species was often the homodimer. A search was therefore made for a method which would place different reactive groups on each protein species such that the rate of interspeties reaction would be substantially faster than the intraspeties reaction. Swan's method for the synthesis of asymmetric disulfides fulfilled these needs (10). RS + R'SSO,-s RSSR' + SO:,"- In the absence of oxygen the initial product is the heteroconjugate. The back reaction can be eliminated with a sulfite trap. As long as base-catalyzed hydrolysis and disulfide exchange can be minimized the heteroconjugate, even though less thermodynamically stable than the homodimer, can be isolated. A further advantage of this reaction is that the product is disul-Reaction 2 involves the Pinner synthesis of methyl-&bromovalerimidate from 5bromovaleryl nitrile. In Reaction 3 the protein amino group is amidinated with BVI" to yield &bromovaleramidinated protein (BVA-protein) which in turn is converted into 5-S-sulfomercaptovaleramidinated protein (SMVA-protein) in Reaction 4 by sodium thiosulfate via the Bunte reaction.
In this communication we wish to present this method of introducing an extrinsic S-sulfonate group into proteins and the ability of the S-sulfonate derivatives of hPL prepared by After 20 min at 4", when the reaction mixture became solidified, the solid was suspended in 20 ml of cold anhydrous ether with the aid of a glass rod and poured (in a hood) on a fritted glass funnel under suction.
The white crystals were washed three times each with 20 ml of cold ether and quickly transferred to a beaker and dried in U~CUO over concentrated sulfuric acid overnight at room temperature.
Twenty-one grams of the product corresponding to 59% overall yield from the starting nitrile was obtained. and stored at -20". Toxin A so prepared usually amounted to 25 to 30 mg and contained <0.5% contamination by intact toxin and toxin B as judged by SDS-disc gel electrophoresis.
The lyophi-lized toxin A was mainly in the disulfide dimer form and was converted into sulfhydryl monomer (toxin ASH) immediately before use as below. Toxin A (10 mg) was dissolved in 0.5 ml of 0.1 M Tris base, 1 mM EDTA, and 6 M urea at pH 9.5 to 10.0. Dithiothreitol was added to a final concentration of 0.1 M and the solution was allowed to stand at room temperature for 15 min. The reduced protein was freed of dithiothreitol by passing through a column of Sephadex G-25 medium (0.9 x 2.3 cm) in nitrogen-purged 20 mM Tes, 1 mM EDTA either in presence or absence of 7 M urea at pH 6.5 and 4". The reduced toxin A so prepared was freed of dimer when fresh, but gradually turned into dimer during storage, especially in the presence of both urea and oxygen.

Preparation
of Peptidyl Elongation Factor 2-A partially purified EF-2 was prepared at 4" from rat liver via "pH 5 enzymes" of Moldave (21) as outlined below. Protein in pH 5 enzymes fraction of rat liver extract from three adult animals was precipitated by adding solid ammonium sulfate to 70% saturation and collected by centrifugation. The pellet was redissolved in 20 ml of 20 rnM TrisiHCl, 1 mM EDTA, 1 rnM dithiothreitol, pH 8.2 (TED buffer) and dialyzed against 50 volumes of the same buffer overnight.
The nondialyzable fraction was clarified by centrifugation and applied on a column of DEAE-Sephadex A-50 (5 x 4 cm) equilibrated in TED buffer. The column was first washed with 50 ml of TED buffer and then eluted with 400 ml each of TED buffer containing 0.07 M and 0.12 M NapSO,, respectively.
Fractions of 20 ml were collected throughout. Protein concentration was monitored by absorbance at 280 nm and activity of EF-2 by toxin A-specific ADP-ribosyl acceptor activity. EF-2 was found to emerge in the protein peak eluted by 0.12 M Na$O,. The EF-2 peak fractions were pooled and adjusted to 70% saturation in ammonium sulfate. After cooling on ice for 15 min, the precipitate was collected by centrifugation and redissolved in 8 ml of TED buffer, dialyzed against 1 liter of the same buffer for 24 h with two changes of buffer. It was then centrifuged to clear off a small amount of precipitate formed during dialysis and stored at -70". EF-2 so prepared usually was 2 to 4 PM in toxin A-specific ADP-ribosyl acceptor and represented a lo-fold purification with 87% recovery of total acceptor activity over the extract. This preparation was also free of ADP-ribosyltransferase activity (22, 23) and stable for several months when assayed under our experimental conditions.

Amidination of Proteins with
Methyl-5-bromoualerimidate-Amidination was usually carried out by adding protein solution of about 60 mg/ml in 20 mM TrislHCl, 1 mM EDTA at pH 8.0 to a weighed amount of BVI. The reagent was dissolved by quick mixing and the pH of the solution was adjusted to the desired pH within 15 s with a few microliters of 1 N NaOH. The solution was then allowed to stand at room temperature for 30 min. At the end of the reaction, 1 N acetic acid was added to bring the pH of amidinated hPL to 6.5 and the sample was freed of hydrolyzed reagent at 4" either by passing through a column of Sephadex G-25 medium (0.9 x 23 cm) in 20 rnM Tes, 6 M urea, pH 6.5, or dialysis against 50 volumes of the same buffer for large scale preparation of amidinated protein.

Reduced and alkylated
RNase (RA-RNase) was amidinated in the same way except 6 M urea was included in the reaction mixture. It was freed of excess reagent by passing through Sephadex G-25 medium column. was then collected on a Whatman GF/C glass fiber filter, washed three times with 3 ml of 5% trichloroacetic acid, three times with 1 ml of 2-propanol/ether, l/2 (v/v), and then twice with 1 ml of 2-propanol/ether/chloroform, 2/2/l (v/v). The dried filter was placed in a counting vial, digested in 1 ml of Nuclear Chicago solubilizer for 20 min at room temperature, and finally mixed with 10 ml of toluene-TLA (Beckman) and counted in a liquid scintillation counter using a narrow "'C window. A zero time control in which toxin A was omitted in the incubation mixture was similarly processed throughout.
One unit of ADP-ribosyltransferase activity was defined as the amount of toxin A or its derivative required to catalyze transfer of 1 nmol of ADP-ribose to EF-2 at 25" in 15 min. For determination of EF-2 concentration the complete system contained 10 fig of toxin A and 0 to 20 ~1 of EF-2 preparation and was incubated for 60 min at room temperature before precipitation with trichloroacetic acid. In each set of gels a number of standard gels of samples of purified toxin A-hPL, ranging from 1 to 10 kg, were also processed with the sample gels under the same conditions.
The wavelength at which each set of gels was scanned was chosen so that the standard gel with highest amount of toxin A-hPL conjugate would give 85 to 90% recorder pen deflection. All gels were scanned at 1.5 cm/min through a 0.2-mm slit with the recorder chart moving at 2 inches/ min. The amount of toxin A-hPL was calculated by integration of the total area under the peak and converted into weight units by interpolation from the linear standard constructed from standard gels.

Amidination of Protein with Methyl-5-bromovalerimidate and Conversion of Product into 5-S-Sulfomercaptovaleramidinated
Protein-The possibility of successful introduction of an extrinsic S-sulfonate group into a protein without disturbing internal disulfide linkage was first tested with reduced and alkylated ribonuclease (RA-RNase) by reacting the protein with &bromovalerimidate, followed by substitution of bromine with thiosulfate.
In the first column in Table I we observe the fall in RA-RNase-free amino groups from 11 to 5 mol/mol after reaction with the bromoalkyl imidate, indicating that extensive amidination has occurred. In the third column -SSO,-groups appear after incubation with thiosulfate. The sum of the remaining free amino groups plus the new -SS03-groups is equal to the original free amino group content (last column).

Synthesis of Diphtheria Toxin A-S-S-Human Placental Lactogen
These data indicate that essentially complete substitution of -Br by -SSO,-has occurred after 2 h of incubation with thiosulfate.
These results show that the 5-bromovalerimidate is a useful reagent for amidination of proteins and the alkyl bromide groups thus introduced are readily converted by thiosulfate to S-sulfonate groups. Substitution-Human placental lactogen was reacted with BVI at various molar ratios and then treated with thiosulfate. The extent of amidination was then determined by assaying the loss of N,PhSO,-reactive amino groups (see Table II) which   Table I, the extent of amidination of the thiosulfate-treated protein is assumed to be equal to the extent of Ssulfonate incorporation.
The contrast between the high ratios of BVI to protein used and the low degrees of amidination obtained indicates that imidate hydrolysis is a major competing reaction. For this reason we have kept protein concentration at high values. It is likely, however, that multiple additions of BVI could serve just as well in cases where protein solubility is a limiting factor.

Reaction of Toxin A-SH with S-Sulfonuted Human
Placental Lactogen-The ability of SMVA-hPL to react with reduced toxin A to form disulfide conjugate was tested by incubation of toxin A and SMVA-hPL together at pH 6.5 and 4" as described under "Experimental Procedures." In Fig. 1

SMVA-bPL
is fully reduced its R, beames identical with that of toxin A, indicating that the more compact structure of the native hPL is due to its internal disultide bonds (30). Toxin A lacks internal disulfide bonds (20) and would be expected to have a more extended structure.
Gel b in Fig. 1 shows a reaction mixture incubated for 48 h before Nethybnaleimide treatment. The new band, Band 2, has an RF in between that of toxin A dimer and hPL dimer, and has a calculated M, = 43,000 as compared to the expected value of 43,372 for a toxin A-hPL conjugate. The individual reacting species are shown in Gels c and d. c contains toxin A dimer (Band 1) and toxin A monmner (Band 4) and d contains bPL dimer (Band 3) and hPL mononer (Band 5).
In these studies different preparations of SMVA-hPL contained varying amounts of dimer. This accounts for the vary ing amounts of the dimer seen in the zero time reaction mixtures. When the reaction was performed with starting materials of low dimer content, a small amount of the dimer was found to be generated.
In Gel a of Fig. 1 a faint band ofR, = 0.342 ? 0.002 is noted. This is present in the SMVA-hPL and is listed as an unknown species in Table III. The R, of this band is clearly distinguishable fmm that ofthe toxin A-bPLconjugate (R, 0.335 k 0.003). The preliminary identification of the reaction product in Fig. 1, Gel b, as toxin A-hPL rests on exclusion that the new band is neither hPL dimer nor toxin A diner, yet has the expected R, value for a conjugate composed of a compact subunit the size of hPL and an extended subunit the size of toxin A.
The ability to react with toxin A-SH to form a disulfide conjugate appears to be a unique property of SMVA-hPL, as is shown by the gels shown in Fig. 2. Thus, when native hPL, BVA-hPL, and SMVA-hPL were independently incubated with toxin A-SH under the same conditions, only SMVA-hPL could react to form the conjugate (Fig. 2, Gel cl. Both native hPL (Gel a) and BVA-hPL (Gel b) could only give rise to toxin A dimer as well as a small amount of unknown species, RF = 0.342, which was also formed in the control reaction mixture of N-ethylmaleimide-treated toxin A with SMVA-hPL (Geld). A similar negative result was also observed with thiosulfatetreated native hPL (not shown).
The reaction of toxin A-SH and SMVA-bPL was also carried out at 25". The yield of conjugate was somewhat less at this FIG. 3. Chromategram of reaction mixture of reduced diphtheria toxin fragment A with S-sulfomereaptovaleramidinated hPL on Sephadex G-150, superfine eolunm.4, reaction mixturesof 9.1 mgof toxin ASH with 30.2 mg of SMVA-hPL (--SSO;/hPL = 0.7 mol/ mall at 4' in a final volume of 3 ml after 120 h, conditions otherwise the same 88 given in Fig. 2. Other experimental methods are given under "Experimental Pmcedures." The orrowed num&rs represent the main band of each peak fraction observed in SDS gels which are in correspondence with the assigned bands shown in Fig. 1 and Table  III. B, reaction mixtures of 13.2 mg of toxin A-8" with 24.8 mg of SMVA-bPL (-8SO;,hPL = 1.8 mal,mol1, diner freed) in a final volume of 1.7 ml after 48 h of incubation under the same conditions as in A, but ADP-ribosyltransferase activity was not assayed. Shaded areas were pooled fractions for purified toxin A-hPL conjugates. Gels of reaction mixtures and pooled fractions are shown in Fig. 4. temperature while the yield of toxin A dimer was slightly greater compared to 4'.
Isolation and Charactertiattin of Toxin A-hPL Conjugate-Isolation of toxin A-hPL conjugate from the reaction mixture was achieved by passing the reaction mixture, without prior treatment withN-ethylmaleimide, through a Sephadex G-150, superfine, column run in 7 M urea. Typical elution profiles of reaction mixtures are shown in Fig. 3, A and B. Protein species are identified by SDS-gel electrophoresis, by absorbance at 280 nm and in addition, in Fig. 3A, by toxin A activity (triangles) assayed as ADP-ribosyltransferase activity. Tbe corresponding reaction mixtures for Fig. 3, A and B, are shown on SDS Gels a of Fig. 4, A and B.
Purified toxin A-WL conjugates obtained by pooling the fractions of Peak 2 marked with black bars are shown in b gels of Fig. 4, A and B.
The reaction mixture shown in Figs. 3.4 and 4.4 contains considerable quantities of toxin A dimer and hPL dimer, rendering the separation less ideal. SMVA-hPL used in Figs. 38 and 4S was freed of hPL dimer by prior gel filtration. These precautions resulted in better peak resolution and increased the purified yield, 19.5% for Fig. 38, as compared to 15.8% for Fig. 3.4, relative to the toxin A input.
The numbxing of the peaks in Fig. 3 corresponds to the numbering used in the gel patterns in Fig. 1 and Table III. In gel filtration in urea, in contrast to SDS gel eleetmphoresis, hPL dimer trails toxin A monmner. This is pmbably due to the different unfolding eff&s of these two reagents (33,34).
The coincidence of protein absorbance of Peak 2 (Fig. 3.4) and ADP-ribosyltransferase activity of Peak 2 should be noted. This coincidence indicates that toxin A activity clearly resides with or without 4 mM dithiothreitol in 20 mM Tes, 1 mM EDTA, pH 7.4, at room temperature for 15 min. N-Ethylmaleimide was then added to a final concentration of 10 mM. An aliquot from each reaction mixture corresponding to 7.4 pg of toxin A-hPL, or 7 pg of toxin A and 6 wg of SMVA-hPL was analyzed in SDS gels. SMVA-hPL used was the one from which the conjugate was prein Peak 2, rather than spilling over from the neighboring toxin A monomer and dimer peaks. In the accompanying paper we show that Peak 2 also displays binding activity toward lactogenie receptors in radioreceptor assays (14). Reduction of the isolated toxin A-hPL conjugate with a high dithiothreitol concentration gave a single band on SDS gels of RF 0.58, a value coinciding with the RF of reduced toxin A and fully reduced hPL. When the conjugate was partially reduced under mild conditions, Fig. 5, Gel b, separation of approximately equal amounts of two subunits was observed. The hPL band in this case had a RF value in between the value for native hPL and fully reduced hPL. An identical R, was observed for SMVA-hPL partially reduced under identical conditions. The three different RF values observed for hPL under varying reducing conditions probably represent reduction of 2, 1, and none of the two disulfide bonds of native hPL (30). Fig. 5 indicates that isolated toxin A-hPL contains approximately equal amounts of toxin A and hPL linked by a disuliide bond.
Estimation of Purity of Isolated Toxin A-hPL Conjugate-The conjugate isolated in Fig. 3B and run in Gel b, Fig. 4B, displays one single narrow band. However, at high loads discernable shoulders in the region of toxin A dimer and hPL dimer could be detected. No monomer contaminants could be detected. Table IV shows the estimated hPL dimer and toxin A dimer contamination determined by integrating scans in these dimer area and taking the ratio of these areas above the baseline to the total peak area above the base-line. The gels scanned contained between 2 and 9 pg of applied conjugate, and conjugate, hPL, and toxin A areas were linear with protein mass over this range. With the conjugate the linear pared. a, toxin A-hPL alone; b, toxin A-hPL + dithiothreitol; c, toxin A plus SMVA-hPL; d same as c but plus dithiothreitol; e, SMVA-hPL + dithiothreitol.
FIG. 6 (right). SDS-acrylamide gels of reaction mixtures containing various initial molar ratios of toxin A-SH and SMVA-hPL incubated for 24 h at 4" under the same conditions as given in Fig. 2. In Gels 1 to 4 toxin A-SH is constant at 0.53 mg/ml and the molar ratio of SMVA-hPL to toxin A is increased as follows: Gel 1, l/l; Gel 2, 2/l; Gel 3, 4/l; Gel 4, 8/l; in Gels 5 to 8, SMVA-hPL is constant at 0.59 mg/ml and the ratio of toxin A to SMVA-hPL is increased as follows: Gel 5, 2/l; Gel 6, 4/l; Gel 7, 7/l. Gel 8 is a zero time control of Gel 4.
The amount of protein electrophoresed is proportional for all reaction mixtures and consisted of 2.93 Fg of toxin A-SH equivalents in Gel 1.
range was determined as low as 0.7 pglgel. In some experiments hPL gave one-third more area per mass on gel scan than toxin A, in others there was no difference. No correction has been made because we find differences of one-third between gel to gel just within the range of error of this method. The contamination of our isolated toxin A-hPL conjugate, Fig.  4, Gel b, appears to be about 5% with hPL dimer and 5% with toxin A dimer on a weight basis.
Effect ofDifferent Molar Ratio of S-Sulfomercaptovaleramidinuted hPL to Reduced Toxin A on Yield of Toxin A-hPL Conjugate-In order to arrive at an optimum condition for preparing toxin A-hPL conjugate in high yield, the effect of varying the molar ratio of SMVA-hPL to toxin A-SH on the yield of the conjugate was studied. The experiments were carried out by incubating a number of reaction mixtures containing a fixed amount of toxin A-SH with varied amount of SMVA-hPL, or vice versa, in the same total final volume. After incubation an aliquot of the reaction mixture was stopped with N-ethyhnaleimide and analyzed by SDS-gel electrophoresis. Typical gel patterns from these experiments are shown in Fig. 6. These data were also treated quantitatively by scanning the gels and integrating the area under each peak. The amount of toxin A-hPL conjugate formed in reaction mixtures of three different preparations of SMVA-hPL (with 0.7, 1.8, and 4.0 S-sulfonate per mol) after 24 (A) and 120 h (B) of incubation are compared in Fig. 7. Thus, when other conditions were kept the same, the yield of toxin A-hPL after 24 h of incubation was always higher when a fixed amount of one reactant was incubated with excess amount of the other (Figs.

S.ynthesis of Diphtheria Toxin A-S-S-Human
Placental Lactogen The per cent contamination of toxin A-hPL conjugate with toxin A dimer and the hPL dimer is shown in the last two columns for five different scanned acrylamide SDS gels of varying loads. This conjugate was isolated in Fig. 3B and is shown in Gel b of Fig. 4. Toxin A and 120 h m B. Lines labeled 2, 3, and 4 connect data points from reactions in which SMVA-hPL is in excess and the horizontal axis is the SMVA-hPL/toxin A ratio. The ratio of -SSO:,-to hPL in moles per mole in SMVA-hPL are as follows; x and 0, 0.7; A, 1.8; 0, 4.0. Line 1 connects points in which toxin A is in excess and the horizontal axis gives the molar ratio of toxin A to SMVA-hPL. All reactions were done in 0.52 ml. At molar ratios of 1, toxin A was 0.53 mgiml and SMVA-hPL was 0.59 mg/ml. Data are from scanned gels. Gels for the points of Lines 1 and 2, Panel A are shown i n Fig. 6. reaction mixtures were compared after 120 h of incubation, the difference in the yield of toxin A-hPL conjugate between low and high molar ratio of the reactants became smaller (Fig.  7B). This observation indicated that the difference in the yield of the conjugate found after 24 h of reaction was mainly due to different rate of formation of the conjugate, as would have been expected from the law of mass action. Upon longer incubation (120 h), when formation of the conjugate reached a plateau, the data became a reflection of the maximum amount of the conjugate obtainable from a fixed amount of one reactant by varying the other. From Fig. 7B, Curve 1, the maximum amount of toxin A-hPL conjugate obtainable from a fixed amount of SMVA-hPL (-SSO,/hPL = 0.7 mol/mol) was found to be 0.14 mg out of 0.28 mg of SMVA-hPL initially present in the reaction mixture. Since the conjugate appeared to be an equimolar mixed disulfide of toxin A and 5-thiovaleramidinated hPL (Table III), it could be calculated from their molecular weights that 48% of the mass of the conjugate was contributed by hPL. Therefore, the above amount of the conjugate found in the reaction mixture corresponded to an overall yield of 24% from SMVA-hPL.
The maximum yield of the conjugate obtainable from an equivalent amount of toxin A-SH (0.31 mg), after reacting with an excess amount of the same SMVA-hPL preparation, appeared to be greater (Fig. 7B, Curve 2) and represented an overall yield of 45%. When the same Toxin A dimer TIME ,h FIG. 8. Time course of product formation for a reaction of toxin A-SH at 0.64 mgiml and SMVA-hPL (-SSO;/hPL = 0.7 mol/mol) at 5.7 mglml at 4". Circles give the toxin A-hPL conjugate and triangles give the toxin A dimer formed. Data are from scanned gels. amount of toxin A-SH was incubated with excess SMVA-hPL preparations with higher S-sulfonate content, the yield of the conjugate became further increased (Fig. 7B, Curves 3 and 4). The overall yield of the conjugate with respect to toxin A was found to be 58 and 53%, respectively, for SMVA-hPL preparations with 1.8 and 4.0 mol of S-sulfonate/mol. Moreover, the rate of reaction of toxin A-SH with these two SMVA-hPL preparations also appeared to be greater than the one containing only 0.7 mol of S-sulfonate/mol (Fig. 7A). These results clearly suggest that for preparation of toxin A-hPL conjugate in high yield, the reaction should be carried out by reacting a fixed amount of toxin A-SH with excess SMVA-hPL containing about 2 mol of S-sulfonate/mol.
When the initial concentration of both reactants was increased 5-fold, the yield of conjugate relative to toxin A input was not significantly altered. Thus with toxin A-SH at 0.63 mglml and SMVA-hPL at 5.7 mg/ml the conjugate concentration after 48 h was 0.34 mg/ml, compared to 1.99 mg/ml, when the concentration was increased 5-fold.

Effect of Urea and Strontium
Chloride Formation of Toxin A-hPL Conjugate-Strontium chloride was included in most reaction mixtures to function as a sulfite trap. Urea was also added to keep protein S-sulfonate in solution in the presence of strontium. These variables were studied. Both urea and strontium ion increased the yield by one-third and also appeared to accelerate the reaction rate. No effect of strontium ion was observed on gel patterns. This was in contrast to earlier studies where the reaction between prolactin S-sulfonate (prepared by sulfitolysis of prolactin) and toxin A-SH was studied. In this study done at a lo-fold higher concentration of reactants, strontium ion was found to be necessary to prevent disproportioning of the prolactin-toxin A product to the homodimers. 1 Time Course of Formation of Toxin A-hPL Conjugate--The time course obtained from scanned gels is shown in Fig. 8. Utilizing 5.7 mg/ml of SMVA-hPL and 0.64 mglml of toxin A-SH as initial concentrations, the half-time of product formation was about 30 min, while 95% completion was achieved at 6 h. The yield of toxin A-hPL in this reacion was approximately 55% relative to the initial toxin A input.

Synthesis of Diphtheria
Toxin A-S-S-Human Placental Lactogen 1513

DISCUSSION
The data presented in the present report clearly indicate that amidination with methyl 5bromovalerimidate followed by nucleophilic substitution of bromine from BVA-protein with thiosulfate provides a convenient route for introducing an extrinsic S-sulfonate group into proteins. The product, S-sulfomercaptovaleramidino-protein, in turn reacts readily with a sulfhydryl-bearing protein, such as reduced diphtheria toxin fragment A (toxin A) to form a cross-linked conjugate in high yield. Dimer and higher polymer formation is negligible. All of these reactions involve mild treatment that is likely to conserve the biologic activity of the proteins cross-linked. Thus amidination of proteins at pH 8.0 involves a mild reaction on primary amino groups with conservation of the positive charge. Retention of biologic activity is usually observed (7,35,36). In fact we have observed retention of biologic activity in BVI-treated hPL (14) with an average number of two amino groups substituted per mol. The subsequent nucleophilic substitution of bromine with thiosulfate appears to be rapid and complete (Table I), as would be expected from the high carbon nucleophilic constant, 6.36, of thiosulfate (37). The S-sulfonated hPL so prepared also retained its biologic activity when assayed in a radioreceptor-binding assay using rabbit mammary gland cell membranes (14).
Some aspects of our reaction of toxin A-SH with S-sulfonated hPL merit discussion. The reaction is actually a reverse of sulfitolysis of a disulfide, as shown by Reaction 1. The K,, of this reversible reaction generally lies considerably to the right in favor of formation of the disulfide. For example, the equilibrium constant for fully ionized cystine is 28 (38). The equilibrium constant of this reaction is generally affected electrostatically by the group(s) adjacent to the disulfide bond in such a way that K,,, is lower for disulfides containing positively charged groups and higher for disulfides containing negatively charged groups (39).
The equilibrium constant for our reaction of toxin A-SH and hPL S-sulfonate is difficult to assess because of the lack of knowledge of the final SOi2-concentration and the probability of heterogeneity of the reactants. When equimolar mixtures of toxin ASH and hPL S-sulfonate were reacted at 0.027 mM, 15 to 20% of the toxin A was found in the toxin A-hPL conjugate (Fig. 7). This figure could be increased to 53% by increasing the hPL S-sulfonate/toxin A ratio. However, little increase of product beyond a 4/l reactant ratio was achieved, indicating that all of the toxin A might not be active. Since 15% of the toxin A was utilized in dimer formation, 30% appears relatively unreactive. Similar calculations from Fig. 7 with excess toxin A indicate that only 24% of the hPL S-sulfonate with 0.7 mol of S-sulfonate/mol is reactive. Because all of the hPL amino groups are potential sites for S-sulfonate introduction, it is likely that steric factors and local electrostatic factors vary the reactivity of the introduced S-sulfonate groups. In the case of toxin A, variable reactivity could result from the charge heterogeneity at the COOH terminus near the reactive cysteine residue. This heterogeneity results from the fact that the toxin A fragment is generated by random tryptic cleavage in a region containing a cluster of 3 arginine residues (2).
We have found no evidence that the toxin A-hPL product disproportions to the symmetric disulfides, as might be expected if a significant reverse attack by sulfite occurred. In Fig. 8 the product appears stable between 24 and 48 h. Although the reaction generates about 15% of toxin A dimer relative to toxin A monomer input, this dimer formation does not appear to result from disproportioning.
The time course fails to show the initial lag characteristic of disproportioning reactions (Fig. 8). Formation of toxin A dimer is independent of the presence of S-sulfonate groups, since toxin A dimer is produced equally well or better in reaction mixtures with native hPL or BVA-hPL (Fig. 2). When hPL or its derivatives are omitted from reaction mixtures, toxin A dimer formation falls to much lower levels. These facts indicate that the oxidizing power for toxin A dimer formation is associated with the hPL preparation. The source of this oxidizing power is unknown but could reside in the internal disulfide bonds of hPL and its derivatives.
Our failure to detect a significant reverse attack by sulfite in the absence of an added SOZ2-trap (Table IV) could be explained in several ways. The backward rate constant could be quite small. For symmetric disulfides this constant varies lOOO-fold depending on the local electrostatic environment (38). The backward reaction could exist and be undetectable if it was assymmetric in character, preferentially generating hPL S-sulfonate rather than a mixture of hPL-SH and hPL Ssulfonate. Such asymmetric nucleophilic attacks are known to occur (40, 41). A third possibility is that our reaction as performed includes a sulfite trap. It is unlikely that at pH 6.5 sulfite is lost as SOB, during the evacuation of air from the reaction vessel. However, it is possible that the internal disulfide bonds in hPL S-sulfonate are more reactive towards sulfite than the S-S bond bridging the toxin A-hPL conjugate. Although the correct possibility is unknown it should be pointed out that we have observed the disproportioning reaction with another conjugate, prolactin-S-S-toxin A, and this disproportioning is effectively reduced in the presence of 100 mM SrCl,, presumably by virtue of the low solubility product of SrSO,, (4 x lo-") calculated from its solubility (42) which leads to a low sulfite concentration (4 x 10e7 M, calculated). The prolactin conjugate was generated by the reaction of prolactin S-sulfonate with toxin A-SH. However, the S-sulfonate groups were not extrinsic, but rather were formed from the internal disulfides by cyclic sulfitolysis and oxidation. ' A number of attempts to make cross-linked protein conjugates either as probes for protein-membrane interactions (43-46) or as agents for selective killing of neoplastic cells (47-52) been recently reported. All these methods involved the mixing of a bifunctional reagent such as glutaraldehyde (43-46, 50, 511, toluene diisocyanate (49, 501, or p,p'-difluoro-m,m'-dinitrophenyl sulfone (48) with the proteins to be cross-linked. Owing to relatively nonspecific reaction of these reagents with amino groups in the proteins, both intramolecular and intermolecular cross-linkages would be formed and among intermolecularly cross-linked conjugates both homo-and heteropolymers would be produced. Purification of an equimolar conjugate, for example ferritin-insulin (46), resulted in very low overall yield (0.02%).
The method of preparing protein conjugates presented in this paper appears superior to methods utilizing bifunctional reagents, since intramolecular cross-linking is prevented and formation of homopolymers is limited. This result leads to high yields of one product, the disulfide-linked conjugate which can then be easily purified. The biochemical and biologic properties of toxin A-hPL conjugates are presented in a companion paper (14).