Artificial Hybrid Protein Containing a Toxic Protein Fragment and a Cell Membrane Receptor-binding Moiety in a Disulfide Conjugate II

The biochemical and biologic properties of a purified disulfide conjugate of diphtheria toxin fragment A and human placental lactogen (toxin A-hPL) have been studied by (a) assaying the ADP-ribosyltransferase activity of the intact conjugate, (b) assaying the binding of the intact conjugate to mammary gland plasma membrane lactogenic receptors, and Cc) assaying the effect of the conjugate on the rate of protein synthesis in rabbit mammary gland explants maintained in organ culture. The toxin A-hPL conjugate retains one-third of the NAD+:EF-2 ADP-ribosyltransferase activity of toxin A, and 26% of the hPL-binding activity to lactogenic receptors. Binding activity was demonstrated by radioreceptor assay and by assaying toxin A activity bound to membranes which was competitively displaced by excess hPL. Since the toxin A-hPL conjugate retained activities of its separate subunits, it could be regarded as a structural analogue of nicked diphtheria toxin with replacement of the original membrane-binding chain by another binding chain that is specific for lactogenic receptor. However, the conjugate failed to inhibit protein synthesis in organ-cultured mammary gland explants, although these were sensitive to native diphtheria toxin and could bind hPL. It is concluded from these results that the toxin A-hPL conjugate does not act as a functional analogue of diphtheria toxin with altered receptor specificity, and that the hPL receptor cannot mediate the entry of toxin A or toxin A-hPL from membranebound conjugate into the cytosol site of action of toxin A.

logue of nicked diphtheria toxin with replacement of the original membrane-binding chain by another binding chain that is specific for lactogenic receptor. However, the conjugate failed to inhibit protein synthesis in organ-cultured mammary gland explants, although these were sensitive to native diphtheria toxin and could bind hPL. It is concluded from these results that the toxin A-hPL conjugate does not act as a functional analogue of diphtheria toxin with altered receptor specificity, and that the hPL receptor cannot mediate the entry of toxin A or toxin A-hPL from membranebound conjugate into the cytosol site of action of toxin A.
A disulfide-linked conjugate of human placental lactogen and diphtheria toxin fragment A has been prepared and purified to 90% purity (1). In this paper we report on some of the biochemical and biologic properties of this artificial protein hybrid conjugate.
The toxin A-hPL' conjugate may be thought of as a struc-* Present address, Unite Institut National de la Santi et de la Recherche MQdicale U34, 29 Chemin Soeur Bouvier. F69322, Lvon Cedex 1, France. _ 1 The abbreviations used are: hPL, human placental lactogen; tural analogue of nicked diphtheria toxin in which the B chain or binding chain of the toxin has been replaced by another binding protein, albeit with different surface membrane receptor specificity (1). The properties of the parent toxin have been well studied (2,3). Nicked diphtheria toxin, a disultide-linked two-chain molecule, binds via its B chain to receptors located on the surface membrane of sensitive cells with an affinity constant of lOa M-I (4,5). After a short lag period the toxin, or its active A chain, enters the cell cytosol through some as yet unknown process. Entry of the toxin A fragment is detected by observing the loss of functional elongation factor II or the loss of protein synthesis which has resulted from ADP-ribosylation of EF-2 catalyzed by the toxin A fragment.
We have synthesized the toxin A-hPL analogue of nicked diphtheria toxin in order to investigate the mechanism of receptor-mediated entry of protein into cells which the parent toxin exhibits. General questions such as the following are of interest: (a) do all surface membrane receptors mediate entry of bound protein molecules, and (b) is the entry of disulfidelinked two-chain toxins directed by only the properties of the binding chain?
In this report we present data showing that both toxin A enzymatic activity and hPL lactogenic receptor-binding activity are conserved in the toxin A-hPL conjugate. However, when toxin A-hPL conjugate is incubated with lactating mammary gland explants, no decrease in protein synthesis occurs, indicating that the lactogenic receptor is unable to mediate entry of the toxin A fragment of the toxin A-hPL conjugate to the toxin site of action within the cell cytosol. These results along with recent findings that lactogenic hormones do enter mammary gland cells (6,7) are discussed in terms of several models of protein entry which are consistent with both sets of data.
toxin A. diDhtheria toxin frament A: toxin A-hPL. a disulfide conjugate oftoxin A and hPL; albumin, dovine serum albumin; EF-2, exists. It is evident that the condition to fulfill this relation is K,IF = K',F' or F' = (K,,/K',,) x F. Thus, from this relation, the difference between the concentration of free conjugate and free prolactin to maintain occupancy of the same number of receptor sites is by a factor ofK,/K', , namely their relative affinity. The relative affinity of prolactin and toxin A-hPL, however, may be closely approximated from the ratio of the concentration of prolactin and toxin A-hPL to exhibit 50% inhibition of maximum specific binding of '*"I-prolactin tracer in the radioreceptor assay (C,,., and C,;., , respectively) as below

Binding of Toxin A-hPL Conjugates and Precursors
to Lactogenic Receptors Determined by Radioreceptor Assay-The specific binding activities of hPL, 5-S-sulfomercaptovaleramidinated hPL and toxin A-hPL conjugates were studied by comparing their ability to competitively inhibit binding of I?prolactin to rabbit mammary plasma membrane. The results are presented by the competition curves shown in Fig. 1. The parallelism of these competition curves suggests that prolactin, hPL, and hPL derivatives all interact with rabbit mammary plasma membrane through the same receptor sites with different binding affinities. The curve for toxin A-hPL conjugate competition is drawn from points (O-----O) obtained from five different preparations. These preparations are similar by this assay and show 26% of the binding activity of unmodified hPL. This activity is more than can be accounted for on the basis of an estimated contamination with hPL dimer of 5% (1,16).
A gradual decrease in binding activity of 5S-sulfomercaptovaleramidinated hPL derivatives was observed when the extent of amidination was increased from about 1 to 4 mol/mol. On the other hand the binding activities of conjugates synthesized from these S-sulfonates were virtually unrelated to the extent of amidination in their precursors. The concentration of total unlabeled hormone or hormone derivatives at which binding of '2"I-prolactin was inhibited by 50% were, respectively, prolactin, 0.325 nM; hPL and SMVA-hPL (0.8), 1 nM; SMVA-hPL (1.81, 1.2 nM; SMVA-hPL (4.01, 2.1 nM and toxin A-hPL conjugates, 3.8 nM.
When the mammary organ culture medium containing 0.3% albumin was substituted for the usual binding buffer, the hPL competition curve for '"Wabeled ovine prolactin was unchanged. However, counts in the membrane pellet were reduced by 50% in this medium due to poor pelleting of the membranes, a consequence of the relative calcium deficiency of the culture medium.
The ability of rabbit mammary explant to bind hPL was also studied by radioreceptor assay using tracer 12"I-prolactin. The results summarized in Table I clearly indicate that rabbit mammary explants are capable of specific binding '""I-prolactin which can be progressively inhibited by increasing concentration of unlabeled hPL. This competition data for explants is comparable with that for membranes shown in Fig. 1. The lower tracer specific binding seen in the explant-binding experiments, 3% as compared to 30% for the membrane experi- ments, is typical of binding studies using explants, and is usually explained by the lower concentration of receptors in tissue fragments and the presence of diffision barriers in explants (17). Nonspecific binding for the explant experiment is 3% using 5 mg of explant (-1 mg of tissue protein) as compared to 2% for the membrane experiment using 80 pg of membrane protein.
Binding of Toxin A-hPL Conjugate to Lactogenic Receptors Determined by Toxin A Activity Assay-The ability of toxin A-hPL conjugates to inhibit binding of 'Wprolactin to rabbit mammary plasma membrane indicates that the conjugate is active in binding to prolactin receptors on the membrane. However, it may be conjectured that the binding activity observed in the above radioreceptor assay is due to hPL or an active derivative generated from the conjugate during incubation with the plasma membrane. In order to eliminate this interpretation of the data, it is necessary to measure directly the binding of intact conjugate to plasma membrane. This may be achieved by measuring the binding of toxin A moiety of the conjugate to plasma membrane in the absence and the presence of excess hPL under the same conditions of radioreceptor assay. If the toxin A moiety of the conjugate is bound to the plasma membrane via specific interaction between the hPL moiety of the conjugate and the lactogenic membrane receptors, toxin A binding should be competitively inhibited by excess hPL. The results of this type of experiment are shown in Table II. At each concentration of toxin A-hPL a certain fraction of toxin A-hPL bound to the membranes is competitively inhibited in the presence of 2.3 PM hPL. This amount of binding is considered to be specific binding and constitutes 49% of the binding at the lowest concentration of toxin A-hPL. The amount of conjugate bound in the presence of 2.3 pM hPL is regarded as nonspecific binding (see "Methods") and is -3% of the total conjugate in each reaction mixture. When toxin A is incubated with membranes, binding also occurs but is unaffected by the presence or absence of excess hPL (not shown). When the data in Table II are plotted as specific bound conjugate against free conjugate, considerable scatter is evident and a binding constant cannot be calculated. In part this is due to the high value of nonspecific binding relative to the specific binding, which must be subtracted from the total binding. It should be emphasized that according to the definition of nonspecific binding (see "Methods" and Ref. 17), the observation of higher nonspecific binding than specific binding is not unusual for an experiment in which high concentration of a weak  Thus, in Fig. 1, the ratio of tracer-specific binding (30%) to nonspecific binding (2%) at 0.122 nM '2SI-prolactin is 15. This ratio is expected to decrease to about 1.5 for a tracer that binds 10 times weaker (toxin A-hPL) if the experiment is carried out at the same tracer concentration (the receptor or plasma membrane concentration are the same for both experiments). This ratio is expected to decrease further with higher tracer concenbinding tracer (in the present case: toxin A-hPL1 is used.
trations which achieve significant fractional occupation of the of Diphtheria Toxin Fragment A-S-S-Human Placental Lactogen 1519 total sites. In order to compare the conjugate-binding experiments utilizing toxin A binding and radioreceptor assays, the expected binding for the Table II experimental conditions has been calculated from the radioreceptor binding curves in Fig. 1 as outlined under "Methods." The found value is about onehalf the expected value at each concentration of conjugate. It appears that at least one half of the conjugate binding observed in the radioreceptor assay represents binding due to intact toxin A-hPL conjugate.

Susceptibility of Interchain Disulfide Bond of Toxin A-hPL and Nicked Diphtheria
Toxin to Reduction-Toxin A-hPL is much more sensitive to interchain disulfide bond reduction than nicked diphtheria toxin. Exposure to 1 mM cysteine, pH 8.2, for 10 min results in about 50% reduction of toxin A-hPL into subunits (Fig. 2, Gel cl, whereas identical conditions applied to nicked diphtheria toxin show no discernible reduction (Fig. 2, Gel j$ Approximately 10% reduction of toxin A-hPL occurs at 0.1 mu cysteine for 10 min (Gel b) .
The buffer and time conditions for these reduction experiments are identical with the assay conditions for toxin A used to assess the ADP-ribosyltransferase activity of toxin A-hPL using dihydroxyethyl disultide-treated EF-2 and N-ethylmaleimide-treated albumin.
The maximum -SH concentration in this assay is 4 pM (see "Methods"). On the basis of the above results no significant reduction of toxin A-hPL conjugate to free toxin A would be expected under this transferase assay.
Almost complete reduction of nicked diphtheria toxin was achieved with 20 mM dithiothreitol (Fig. 2, Gel g). This reagent was subsequently used to assess the toxin A content of a bcdefg conjugates by their ADP-ribosyltransferase activity following reduction.
Content of Toxin A in Toxin A-hPL Conjugates-The content of toxin A in all three toxin A-hPL conjugates measured in the presence of 40 mM dithiothreitol is summarized in Table  III and is equal to about one-half the conjugate mass. These values agree with the values calculated for a 1:l toxin A-hPL conjugate, considering the 10% error of the assay (see "Methods"). A control measurement of the toxin A content of nicked diphtheria toxin is also close to the expected value. These results are also in agreement with the previously reported results of SDS gel electrophoresis of partially reduced conjugate (1).
ADP-ribosyltransferase Activity of Toxin A-hPL -The toxin A moiety of toxin A-hPL conjugate retained part of its ADP-ribosyltransferase activity when the intact conjugate was assayed in the absence of dithiothreitol using dihydroxyethyl disulfide-treated EF-2. As summarized in Table IV, all but one of the assayed toxin A-hPL preparations were about one-third as active as free toxin A-SH in ADP-ribosyltransferase activity on an equimolar basis. The lower activity found in toxin A-hPL prepared from SMVA-hPL containing four introduced -SS03-groups may reflect the effect of this substituent on the toxin A activity.
Since the conditions of this assay are not expected to reduce the conjugate to subunits (see previous two sections) these results are believed to reflect the activity of the intact conjugate. Evidently, the enzymatic site within the conjugate is exposed in contrast to the situation existing in unreduced nicked diphtheria toxin (3). This result is not surprising since mu-educed CRM45, an early termination mutant of diphtheria toxin lacking the terminal portion of the B chain, also shows considerable transferase activity (2).
Since the transferase activity of the conjugate relative to equimolar toxin A is one-third under nonreducing conditions (Table IV) and rises to 100% under reducing conditions (Table   TABLE   III Toxin A content of toxin A-hPL conjugates and nicked diphtheria toxin Toxin A content is found from the ADP-ribosylation activity in the presence of 40 mM dithiothreitol as described under "Methods." Calculated values are based on subunit mass ratios given in the preceding paper (1). Nicked toxin was prepared as previously described (1   III) it was of some interest to study the range of intermediate reducing conditions. This was done by preincubating conjugate or nicked diphtheria toxin with various concentrations of cysteine and then measuring ADP-ribosylation after a 1250fold dilution (in order to minimize varying thiol effects on the assay). These results are shown in Table V and should be compared with the cysteine concentrations required for reduction of the interchain disuliide bond in Fig. 2. Nicked toxin activity is correlated with interchain disulfide reduction, as has been previously reported (3). Conjugate activity is 46% of the maximum in the absence of added cysteine. The cysteine concentration at which one-half of the increase to maximum activity occurs is 1 mM, a concentration found by the gel data to give 50% reduction.

Mammary
Gland Organ Culture Assay of Toxin A-hPL; Effect on Protein Synthesis- The effect of toxin A-hPL conjugate and its precursors on the rate of protein synthesis of lactating rabbit mammary gland explants are shown in Table  VI. After 20 h of incubation the counts from L-[YJleucine incorporation into protein are the same for explants exposed or not exposed to toxin A-hPL conjugate. The presence of hPL resulted in a 26% increase in incorporation over no addition, which may be significant.
Intact diphtheria toxin causes a reduction of over 95% in incorporated counts while the counts in the toxin A addition is down by 80%. This latter effect is undoubtedly due to the intact or nicked toxin contamination of the toxin A which we estimated by SDS gel electrophoresis to be 0.3%. It appears that this assay at 20 h is sensitive to 15 ng/ ml of intact toxin, a range of sensitivity equal to that of HeLa cells in tissue culture (18). After 4 h of incubation the effect of 5 pg/ml of diphtheria toxin is already seen with a 70% reduction in L-[Wlleucine incorporation; however, the lower concentration of toxin in the toxin A control has not had sufficient time to achieve its potential inhibition.
At the end of both the 4-and 20-h incubations of toxin A-hPL conjugate with explants, the presence of intact conjugate was documented by both SDS gel electrophoresis and radioreceptor assay of the culture medium. No appreciable loss of conjugate was detected. These results show that toxin A-hPL conjugate is without detectable effects on mammary gland protein synthesis under conditions where (a) the gland is sensitive to the effects of toxin and (b) the gland contains hPL receptors.

DISCUSSION
The results presented in this report show that the toxin A-hPL disuliide conjugate which we have synthesized and purified retains both lactogenic receptor-binding activity of hPL and ADP-ribosyltransferase activity of toxin A. Transferase activity was assayed in the absence of reducing agent using an EF-2 preparation treated with dihydroxyethyl disulfide. The assayed activity was 33% of the activity found after reduction of the interchain disulfide bond. This level of activity most likely reflects the activity of the intact conjugate and cannot be accounted for by the 5% estimated toxin A dimer contamination of the conjugate (1). In contrast to the conjugate, nicked diphtheria toxin has only 1% of its full activity when assayed under nonreducing conditions. The masking effect is associated with the COOH-terminal portion of the B chain and is absent in the toxin mutant CRM45 which is missing this region (2).
Lactogenic binding activity of the conjugate was assayed utilizing lactating rabbit mammary gland membranes and ?-labeled ovine prolactin tracer. The competition curves for prolactin, hPL, and toxin A-hPL conjugate are parallel, but displaced toward higher concentrations, indicating that the same receptor is involved in each case but with altered affinities (Fig. 1). Toxin A-hPL conjugate was 26% as active as hPL on a molar basis. This level of activity could not be explained by the 5% estimated contamination of conjugate with hPL dimer (1). The reduced binding activity of toxin A-hPL is not unexpected, since the additional protein mass would more than likely interfere with the receptor-hPL conformational fit which has presumably been selected by evolutionary pressure. An example of extra protein mass reducing binding activity is proinsulin, where binding to insulin receptors is reduced by 80% as compared to insulin (19). To eliminate the possibility that the lactogenic binding activity observed in the radioreceptor assay of conjugate was due to free hPL, split from the conjugate during assay, the binding of conjugate was followed by a toxin A assay. Specific binding of conjugate via lactogenic receptors was considered to be the binding which was effectively blocked by excess of free hPL. When conjugate at 0.33 nM was incubated with membranes, 0.047 pmol of toxin A/mg of membrane protein was bound. When the incubation was performed in the presence of excess hPL this value dropped to of Diphtheria

Toxin Fragment
A-S-S-Human Placental Lactogen 1521 0.023 pmol/mg giving a ratio of specific binding to total binding of 49% (Table II). The amount of conjugate bound at each' conjugate concentration was about one-half the value calculated from the radioreceptor competition curves. However, because of scatter in the points of the radioreceptor assay the exact position of the toxin A-hPL conjugate competition curve is uncertain, and the observed and calculated values lie within this uncertainty.
Since the toxin A-hPL conjugate contained both toxin A activity and lactogenic receptor-binding activity it was of some interest to see whether this conjugate could act as a functional analogue of diphtheria toxin with altered receptor specificity. Could the lactogenic receptor mediate the entry of toxin A from bound toxin A-hPL conjugate to its site of action in the cytosol? This was tested by incubating lactating rabbit mammary gland explants with conjugates and assaying protein synthesis. No effect was observed under assay conditions which were sensitive to concentrations of intact diphtheria toxin 330-fold less than used for toxin A-hPL (Table VI). At the end of the incubation of explant with conjugate, the presence of conjugate was documented by assaying the organ culture medium by gel electrophoresis and radioreceptor assay. In addition, the activity of the lactating mammary gland explants used in this experiment to bind lactogenic hormones was documented by radioreceptor assay. These results indicate that the lactogenic receptor is unable to mediate the entry of toxin A or toxin A-hPL from the receptor-bound conjugate to the cell cytosol. Several explanations of these results are possible. The conjugate could differ in important properties from nicked diphtheria toxin and therefore not represent a functional analogue. The enhanced susceptibility of the conjugate to reduction compared to nicked toxin might result in interchain reduction and chain separation at the cell surface prior to entry. Another possibility is that the toxin A moiety, while not appreciably interfering with conjugate binding, might sterically interfere with entry. Experiments to test these possibilities are in progress.
Other explanations of these results are concerned with the properties of the lactogenic receptors. A simple explanation states that there are two classes of receptors, those specialized for protein entry into the cytosol and those lacking this characteristic, and lactogenic receptors fall into the latter category. This explanation, however, must be reconciled with recent data which indicate that lactogenic hormones enter the interior of mammary gland cells. Whitworth and Grosvenor (6) have reported that during an intravenous infusion of prolactin into lactating rats the prolactin concentration in the milk rose and reached a steady state value of 250 nglml, compared to the plasma steady state value of 230 nglml. After terminating the infusion, plasma prolactin levels fell rapidly to the preinfusion level but milk prolactin fell more slowly. Nolin and Witorsch (7) have reported the presence of prolactin within the alveolar cells and milk-containing ducts of lactating rats using histologic techniques dependent on a reaction involving antiprolactin antibodies and a second chromogenic antibody. These two studies indicate that prolactin can enter mammary cells from the circulation and be transported across these cells into the milk ducts. It is not known whether this process is receptormediated or whether the intracellular prolactin is free within the alveolar cell cytosol or enveloped in a membraneous vesicle. Transport of proteins across cells has been previously studied, most extensively for cases where maternal antibodies cross into the fetal circulation. These transport mechanisms have a large capacity and specificity is present (20). Receptors for the Fc fragment of y-globulin are probably involved, since Fc will compete with y-globulin for transport. Evidence for compartmentalization during transport has been found. The entry of diphtheria toxin into sensitive cells can proceed by mechanisms which do not lead to loss of protein synthesis (21). Using biologically active ""I-labeled diphtheria toxin, cellular uptake can be quantitated.
Uptake is rapid, involves large quantities of toxin and is not detectably influenced by the presence of 4 mM ammonium ion. However, this concentration of NH,+ prevents toxin-mediated loss of protein synthesis. Ammonium ion is without effect on toxin binding to cells or on the ADP-ribosylation of EF-2 in vitro. The level of toxin uptake determined by labeled toxin is either in the presence or absence of NH,+, over lOOO-fold more than that required to inactivate protein synthesis. Using the turnover number of toxin A, the intracellular EF-2 concentration and rate of synthesis of EF-2, it has been calculated that only a few toxin molecules per cell are needed to inactivate protein synthesis (2). One interpretation of these results, proposed by Bonventre et al. (211, which we find attractive is that there are two entry mechanisms which distribute toxin (or other proteins) to two separate sites. The mechanism observed with labeled toxin is a high capacity system and the entered toxin is bound within endocytotic vesicles or some other compartment from which it cannot escape. This mechanism may be similar to the mechanism for prolactin transport and maternal globulin transport into fetal circulations. The second mechanism is a low capacity system receptor-mediated, transporting at most only 100 to 1000 molecules/h to the soluble phase of the cell cytosol. This mechanism is responsible for action of diphtheria toxin and is blocked by NH,+. Toxin or conjugates containing toxin A transported by the first mechanism would be inactive, since they could not interact with EF-2 which cycles between the soluble phase of the cell cytosol and its ribosomal binding site.
In summary, a conjugate protein hybrid of diphtheria toxin Fragment A and human placental lactogen has been synthesized and purified. This conjugate retains the separate binding and enzymatic activities of its subunits. For use as a probe the conjugate can be regarded as a structural analogue of diphtheria toxin in which the receptor-binding chain of the toxin has been substituted by another binding protein, now made specific for lactogenic receptors. When assayed with mammary gland cells containing lactogenic receptors no intracellular toxin activity was detected, indicating that the toxin A-hPL conjugate did not behave as a functional analogue of diphtheria toxin. This occurred in spite of evidence indicating that lactogenic hormones can enter and be transported through mammary gland cells. Several explanations focusing on both the properties of the receptors involved and the properties of the toxin analogue which we have synthesized have been considered. Whether or not other analogues of diphtheria toxin can be synthesized which have altered receptor speciflcities and which can function to inhibit cellular protein synthesis will require further investigation.