Ligand-toxin Hybrids Directed to the (cid:97) 2 -Macroglobulin Receptor/ Low Density Lipoprotein Receptor-related Protein Exhibit Lower Toxicity than Native Pseudomonas Exotoxin*

Pseudomonas exotoxin (PE) binds the heavy chain of the (cid:97) 2 -macroglobulin receptor/low density lipoprotein receptor-related protein (LRP). To understand the sig-nificance of this interaction, novel toxin-derived gene fusions were constructed with two ligands that also bind this receptor. A 39-kDa cellular protein, termed RAP, binds LRP with high affinity and often co-purifies with it. Two RAP toxins were constructed, one with PE and one with diphtheria toxin (DT). RAP, which replaced the toxins binding domains, was combined with each of the corresponding translocating and ADP-ribosylating domains. Both RAP-toxins bound LRP with an apparent higher affinity than native PE. Despite this, RAP-PE and DT-RAP were less toxic than native PE. Apparently, RAP-toxin molecules bound and entered cells but used a pathway that afforded only low efficiency of toxin trans- port to the cytosol. This was evident because co-inter-nalization with adenovirus increased the toxicity of RAP-toxins by 10-fold. We speculate that the high affinity of RAP binding may not allow the toxin’s translocat- ing and ADP-ribosylating domains to reach the cytosol but rather causes the toxin to take another pathway, possibly one that leads to lysosomes. To test this hypoth-esis, additional RAP-PE fusions were constructed. N- terminal or C-terminal fragments of RAP were joined to PE to produce two novel fusion proteins

The ␣ 2 -macroglobulin receptor/low density lipoprotein receptor-related protein (LRP) 1 is one of the largest membraneassociated proteins characterized to date. Its primary amino acid sequence was derived from overlapping cDNA clones (1).
LRP is located both on surface and intracellular membranes of many eukaryotic cell types. It is synthesized as a 4525-amino acid single chain precursor. After synthesis, possibly in the Golgi, single chain LRP is cleaved to give a 515-kDa heavy chain and an 85-kDa light chain (2). The light chain contains both a membrane-spanning region and cytoplasmic tail. The heavy chain which remains membrane-associated via non-covalent interactions with the light chain, contains the ligandbinding sites. LRP mediates the binding and endocytosis of several unrelated ligands: 1) apoprotein E-enriched ␤-migrating very low density lipoprotein (3,4); 2) protease-or methylamine-activated ␣ 2 -macroglobulin (5); 3) the complex between tissue-type plasminogen activator (tPA) or urokinase with plasminogen activator inhibitor type 1 (PAI-1) (6,7). A 39-kDa receptor-associated protein, called RAP, also binds to LRP (8,9). Recently, it was shown that RAP has two regions that can independently, although less efficiently, interact with LRP and another protein from the same family: gp330 (10,11,12). RAP inhibits the binding and or uptake of all known ligands that interact with LRP. At a minimum, RAP binds LRP with high affinity at two sites and possibly causes a conformational change in receptor structure. The addition of either heparin or ␤-mercaptoethanol inhibits the interaction of RAP with LRP (13). Recently, LRP was proposed as the cell surface receptor for Pseudomonas exotoxin (PE) (14 -16).
Both PE and diphtheria toxin (DT) inactivate the protein synthetic apparatus of eukaryotic cells in a series of steps that include: binding to a surface receptor, endocytic uptake, cellmediated proteolytic processing to generate an enzymatically active fragment, translocation of the fragment to the cell cytosol, and ADP-ribosylation of elongation factor 2.
PE is synthesized as a single chain bacterial protein composed of three structural domains. The N-terminal domain mediates binding to LRP: specifically, Lys-57 and possibly other nearby residues mediate binding (17). Toxicity for cells is reduced by at least 100-fold when Lys-57 is changed to glutamic acid or when most or all of this domain is deleted. The middle domain of PE has two functions: it contains sequences necessary for translocation to the cell cytosol and it serves as a substrate for cell-mediated cleavage (18,19). The C-terminal domain has the ADP-ribosylating activity and contains a putative endoplasmic reticulum (ER) retention sequence (20 -22). Functionally DT is quite similar to PE. However, the binding and ADP-ribosylating domains of DT are located in the opposite orientation to PE, i.e. the binding domain of DT is at the C terminus (23) while the ADP-ribosylating activity is at the N terminus (24). Like PE, the middle domain of DT contains sequences that mediate translocation to the cell cytosol (25). While the exact intracellular location for the translocation of either toxin has not been clearly defined, existing data suggests that the A chain of DT reaches the cytosol from an acidic endosomal compartment while the corresponding PE fragment needs to reach the ER to facilitate its translocation. Thus it appears that these functionally similar toxins use different intracellular pathways to reach the same cytosolic location.
By replacing the binding domains of either PE or DT with binding ligands of various specificities it has been possible to redirect toxic activity to cells bearing particular cell surface receptors (26). For instance, DT and PE have been targeted to IL-2, IL-4, IL-6, and epidermal growth factor receptors. And although some losses in binding affinities has been noted, it is often possible to produce active hybrid toxins by placing a particular ligand either at the N terminus of PE or the C terminus of DT.
As a way to study both intracellular receptor traffic and possible differences in toxin processing, we have constructed a number of ligand-toxin hybrids which bind to LRP. Because ligands will interact with different portions of the same receptor and possibly bind with different affinities, it may be possible to determine which interactions lead to efficient delivery of active toxin fragments to the cell cytosol and which do not.
To create hybrid molecules with cytotoxic characteristics that can be compared with native PE, we have replaced the receptor-binding domains of DT and PE with either RAP or tPA-PAI-1. Results indicated that these hybrid proteins exhibited much lower cytotoxicity than native PE. We also produced RAP-PE fusions composed of RAP fragments in place of fulllength RAP. These proteins were more active than toxin fusions made with whole RAP. Based on these data we speculate that RAP-and tPA-PAI-1-containing recombinant proteins possess lower cytotoxicity than PE because they are less capable of dissociating from LRP and therefore are transported to lysosomes more readily than PE.
Oligonucleotides-The following oligonucleotides were from Bio-Serve Biotechnologies and were used for polymerase chain reaction amplification: 5Ј-TGCAGGTCGGCGGCCGCCTCGCGGGAGAAGAA-CCAGC (rap-NotI) and 5Ј-TTTTTTTTTTGGTACCCCGAGTTCGTT-GTGCCGAGCTCT (rap-KpnI) were used to amplify a cDNA sequence encoding whole RAP, using pRAP1-23 as a template; while rap-NotI and 5Ј-TTTTTTTTTTTTGGTACCAGGTCCGAGGGGCTAATGAC (RAP-KpnI) were used to amplify a cDNA sequence encoding the Nterminal portion of RAP (amino acids from 1 to 175) and 5Ј-AAAA-AAAAAAAAGCTTCAGCGGCCGCGGACCTGAGCGACATCAAGG (RAP-HindIII) together with rap-KpnI were used to amplify a sequence encoding the C-terminal portion of RAP (amino acids from 172 to 323). 5Ј-TTGGTGAAGGGGATCTGTGCACCATCCCCCATCCT and 5Ј-CCC-AGGGTCAGGGTTCCATCACTTGGCCCAT were used to amplify the sequence encoding PAI-1.
Reagents-tPA was a gift from Dr. Uli Brinkmann, NIH. N-Methylsulfonyl-D-Phe-Gly-Arg-4-nitroanilide acetate was from Boehringer Mannheim. Polyclonal rabbit antibodies against the heavy chain of LRP were prepared as described (12), polyclonal horse antibodies against DT conjugated with horseradish peroxidase were a gift from Dr. Smirnov (Russia). Human liver poly(A) RNA was from Clontech. Heparin was from Sigma. Heparinase was from U. S. Biochemical Corp.
Expression and Purification of Recombinant Proteins-At an absorbance of 0.5-0.6 (590 nm), protein expression was induced by the addition of isopropyl-␤-D-thiogalactoside. Cells were harvested 90 min later. DT-RAP was recovered from the periplasm of E. coli by osmotic lysis. Although RAP-PE hybrids were found in a soluble form in the cytoplasm, in the presence of rifampicin most of the proteins were insoluble. Therefore these proteins as well as DT-PAI-1 were obtained after dissolving inclusion bodies in 7 M guanidine hydrochloride and renaturing in 10 mM Tris-HCl, 1 mM EDTA, 300 mM arginine, pH 6.5. Proteins were further purified using successive rounds of ion exchange chromatography. During purification of some the RAP-containing proteins, an affinity chromatography step with heparin-Sepharose (Pharmacia) was also used.
Tissue Culture-Murine L-929 cells, green monkey COS cells, and human A431 cells were obtained from ATCC. Chinese hamster ovary line, CL6, and mutants 221-1 and 13-5-1 were from Dr. S. Leppla (NIH). All lines were maintained in RPMI 1640 medium, 5% fetal bovine serum with penicillin and streptomycin. Sf9 cells, derived from Spodoptera frugiperda pupal ovarian tissue were from Dr. J. T. Schiller (NIH). SL2 cells, derived from Drosophila melanogaster, were from Dr. I. Krasnoselskaja (NIH). Insect cells were maintained in Bac-V medium from Stratagene supplemented with 5% fetal bovine serum.
Cytotoxicity Assay-Cells at 1 ϫ 10 5 per well (24-well plates) were seeded 1 day prior to evaluating the cytotoxicity of the toxin-related proteins. Various concentrations of toxins and ligand-toxins were added to cells for 20 h at 37°C. At the end of this period, [ 3 H]leucine was added to a final specific activity of 2 Ci/ml for a further 1 h. To remove unincorporated radioactivity, monolayers were washed with phosphatebuffered saline. Proteins were then precipitated with trichloroacetic acid and the radioactivity per well determined by solubilizing the monolayers with 0.1 N NaOH and counting in a liquid scintillation counter. The results are expressed as percentage of control wells to which no toxin was added.
To determine the effect of adenovirus on cytotoxicity of RAP-PE, 96-well plates were used. Cells at 4 ϫ 10 4 per well (96-well plates) were seeded 1 day prior to evaluating the cytotoxicity. Various concentrations of toxins and RAP-PE were added to cells for 2 h with or without non-toxic amounts of adenovirus type 2 (approximately 1 g/ml of virus) at 37°C. At the end of this period, the existing medium was substituted by fresh medium containing [ 3 H]leucine (0.5 Ci/ml) and cells were incubated further at 37°C for another 1 h.
Competition Assay-PE at 300 ng/ml, in the presence or absence of potential competitors, was added to cells for 4 h at 37°C. During the last hour of the 4-h incubation, [ 3 H]leucine was added to a final specific activity of 2 Ci/ml. To determine the extent of inhibition of protein synthesis, the monolayers were processed as described above.
Receptor Binding Studies-LRP from bovine liver was affinity purified on a DT-RAP column, as described (32). Different amounts of purified receptor were applied to individual wells of an Immulon microtiter plate. Following the addition of bovine serum albumin to block nonspecific binding, PE or DT-RAP with or without competitors were added to the receptor-coated wells. Evidence of PE and DT-RAP binding was obtained by adding peroxidase-labeled (Jackson Laboratories) affinity-purified rabbit anti-PE or peroxidase-labeled affinity-purified horse anti-DT.

Construction and Expression of RAP-toxin Gene Fusions-A
polymerase chain reaction fragment encoding full-length RAP was generated from pRAP1-23. Subsequently, RAP-PE and DT-RAP gene fusions were constructed, as illustrated in Figs. 1 and 2. In the RAP-PE construct (pET39PA2), RAP replaced amino acids 1-103 at the N terminus of PE. For the generation of DT-RAP (pVCDT39-1), RAP was positioned on the C-terminal side of amino acids 1-388 of DT. Expression levels in E. coli of both hybrid toxins ranged from 20 to 30% of the total cell protein. The DT-RAP fusion protein accumulated in the periplasm, while the RAP-PE hybrid was found mostly in the cell cytoplasm as a soluble protein. Since the DT-RAP fusion lacked a signal sequence, the observed secretion to the periplasm had not been expected.
Properties of RAP-PE and DT-RAP-To determine if the two RAP-toxin fusions retained binding activity, a series of binding and ligand blocking experiments were devised. Since native RAP binds heparin, we first assessed the heparin-binding ca-pacity of both RAP-PE and DT-RAP. Both hybrid proteins bound to a heparin-Sepharose column and both were eluted with the application of a linear gradient of sodium chloride: RAP-toxins eluted between 0.8 and 1.0 M NaCl (data not shown). Since similar conditions are used to elute native RAP, we concluded that the heparin binding activity of RAP was unaffected by a fusion to either the N terminus of PE or the C terminus of DT.
Next we investigated whether or not the RAP fusions retained their ability to interact with LRP. Results obtained with an enzyme-linked immunosorbent assay indicated that RAP-PE bound to LRP at both pH 5.5 and 7.2, and that the addition of heparin (9 g/ml) as a competitor reduced this binding to very low levels (Fig. 3). Previously, Moestrup and Gliemann (13) reported that the addition of heparin completely abolished the binding of RAP to LRP. Furthermore, when we preincubated heparin with heparinase its blocking activity was completely lost, indicating that specific sequences of monosaccharides within heparin were responsible for the inhibition. Using the same assay conditions, PE was shown to bind to LRP but with a much lower apparent affinity (Fig. 3). PE binding was not inhibited by heparin. In ligand blots following SDS-PAGE run under non-reducing conditions, DT-RAP was seen to interact with the heavy chain of LRP (data not shown). This binding was abolished when LRP was treated with ␤-mercaptoethanol. As a further indication that DT-RAP retained LRP binding activity, we report that immobilized DT-RAP could be used to affinity purify LRP from beef liver (data not presented).

Toxicity of Ligand-toxin Hybrids Directed to LRP
Previously, it was shown that the addition of RAP to intact cells, blocked PE binding and PE-mediated cytotoxicity (14). In Fig. 4 we show that the addition of DT-RAP inhibited cytotoxic activity of PE for L929 cells. Significant reversal of PE toxicity was seen when DT-RAP was added in the range of 11-33 nM. (At these concentrations, DT-RAP by itself, did not inhibit protein synthesis.) Similar RAP concentrations are known to block the binding of physiologic ligands.
We also found that the fusion of PE with RAP did not alter the susceptibility of PE to furin-mediated cleavage (Fig. 5, arrows, indicate the generation of proteolytic fragments with time of incubation).
Cytotoxicity of DT-RAP and RAP-PE-LRP is widely expressed in a variety of mammalian cells and tissues. It is also found in cells of Caenorhabditis elegans (33). These data suggest a wide distribution of this receptor in animals. To determine the cytotoxicity of RAP-PE and DT-RAP, relative to native PE, RAP-toxins were added to cell lines from five different species. These included: murine L929 cells, a line which is very sensitive to PE; COS cells, a green monkey kidney line, which is very sensitive to DT; A431, a human epithelial cell line; Chinese hamster ovary cells; and Sf9 insect cells. Because the RAP-toxins bound to LRP with higher affinity than native PE, we anticipated that they might exhibit greater toxicity than PE itself. Based on previous results, this could be expected since in six different epidermal growth factor receptor-positive cell lines, TGF␣PE40 was generally 10-fold more toxic than native PE (34). Surprisingly, the reverse was seen with the RAPtoxins. In all lines tested, PE was more active than either RAP-PE or DT-RAP (Table I). In the two lines most sensitive to PE, L929 and COS, native PE was 80 -500-fold more active than the hybrid toxins. These differences were present but less pronounced in the other three lines.
To determine if RAP toxin was being internalized to the endosomal compartment, cells were co-incubated for 2 h with RAP-PE and non-toxic amounts of adenovirus type 2 (approximately 1 g/ml of virus). Previously, we showed that adenovi-rus disrupts endosomal membranes and releases the contents to the cytosol (35). Results indicated that when adenovirus and RAP-PE were added to cells together, cytotoxicity was enhanced by approximately 5-10-fold (Fig. 6). From this we conclude that RAP-PE binds and enters cells but is less toxic than the native toxin because the toxin portion of the hybrid is transported to a location which does not allow efficient translocation to the cytosol.
To examine possible reasons for the relatively low toxicity of the RAP-toxins, additional experiments were performed on three cell lines all having the same genetic background. Wild type (WT) Chinese hamster ovary cells (clone CL6) and two lines characterized as PE-resistant (13-5-1) and PE-supersensitive (221-1) were used. The 13-5-1 cells have no detectable LRP (16) while the 221-1 line appears to express slightly higher levels of LRP than wild type. Compared to WT, the LRP-  5 and 13), 18 h (lanes 6 and 14), and 20 h (lanes 7 and 15) followed by analysis on SDS-PAGE and staining with Coomassie Blue. Numbers correspond to molecular masses of protein standards (lane 10). Arrows indicate the position of PE and RAP-PE fragments generated by furin.
negative line was 100-fold resistant to PE, while the 221-1 was 3-fold more sensitive (Fig. 7). On WT and 221-1 cells RAP-PE was 5-and 10-fold less active than native PE, respectively (Fig.  7). While 13-5-1 cells were 10-fold resistant to RAP-PE compared to WT, the absolute activity of RAP-PE and PE for this cell line was quite similar. DT-RAP had about the same activity as RAP-PE on WT and 221-1 cells (Fig. 7). However, the 13-5-1 cells exhibited only slight resistance to DT-RAP. Thus, in the context of a RAP fusion, PE toxicity was much more dependent on delivery by LRP than was DT. Since DT can translocate from the endosomal compartment and PE must reach the endoplasmic reticulum, it is possible that internalization on any receptor will allow efficient delivery of DT to the cytosol but not PE (see "Discussion").

Construction and Cytotoxicity of RAP-PE Hybrids Composed of N-terminal and C-terminal RAP Fragments-
To study the relationship between binding to LRP and cytotoxicity, we decided to take advantage of recently published data reporting that the N-and C-terminal portions of RAP can bind independently but less efficiently than full-length RAP to LRP and to the protein from the same family: gp330 (10 -12). Therefore by replacing a small NotI-KpnI fragment in the plasmid pET39PA2 with NotI-KpnI fragments amplified from plasmid pET39-5 using either rap-NotI and RAP-KpnI or RAP-HindIII and rap-KpnI (see "Materials and Methods") primers we constructed plasmids pETN39PA5 and pET39CPA11. These plasmids encode RAP-PE hybrids that are composed of the N-and C-terminal portions of RAP, respectively. When the cytotoxic activities of these proteins were analyzed, the RAP fragmenttoxin hybrids were found to be 5-10-fold more toxic than whole RAP-PE (Fig. 8).
Construction and Expression of a DT-PAI-1 Gene Fusion-To study the fate of another ligand that binds LRP, we constructed a fusion between DT and the protease inhibitor PAI-1 (Fig. 2).
Polymerase chain reaction was used to amplify the appropriate PAI-1 sequence (36) from total human cDNA. Primers were designed to produce a PAI-1 construct that lacked a signal sequence. The resulting plasmid called pVCDT-PAI-1-4 was transformed into BL21(DE3) cells and the hybrid protein (here called DT-PAI-1) was expressed. The expression level reached approximately 20% of the total cell protein. DT-PAI-1 had the expected molecular mass of 97.5 kDa and was recognized by horse antibodies to DT. Because it was insoluble, DT-PAI-1 was recovered from inclusion bodies by denaturation with guanidine hydrochloride and renaturated at either pH 6.5  PE  10  260  38  115  150  RAP-PE  1,800  1,500  7,000  13,000  7,000  DT-RAP  800  1,300  20,000  or 8.0. Although similar yields of soluble protein were achieved at either pH, the DT-PAI-1 renatured at the higher pH interacted with tPA less well than the hybrid renatured at pH 6.5. PAI-1 can exist in either an active or a latent (inactive) form (37). Over time, the active form converts into the latent form. Apparently, this conversion proceeds faster at pH 8.0 than at 6.5 (38).
Interaction of DT-PAI-1 with tPA-The interaction of PAI-1 with tPA produces a complex which is stable to treatment with SDS. To determine if DT-PAI-1 could interact with tPA in a similar fashion, the two were mixed and the formation of a stable complex analyzed by SDS-PAGE. Results indicated that a novel molecule of molecular mass 160 kDa was visualized (Fig. 9). In Western blot analysis, this complex reacted with antibodies to DT. To determine if DT-PAI-1 exhibited protease inhibitory activity, it was added to a reaction mixture comprised of tPA and the substrate N-methylsulfonyl-D-Phe-Gly-Arg-4-nitroanilide acetate. In the presence of DT-PAI-1, there was dramatic inhibition of substrate hydrolysis (data not presented).
Cytotoxicity of DT-PAI-1-tPA Complex-Once formed, DT-PAI-1-tPA should bind to the heavy chain of LRP. To determine if this complex was toxic for cells, the complex or DT-PAI-1 alone was added to either COS or A431 cells. There was no toxicity for A431 cells (data not shown). However, the complex was toxic for COS cells (Fig. 10). DT-PAI-1-tPA, which had low specific activity compared to either PE or DT, was about 16 times more toxic for COS cells than DT-PAI-1 alone. DISCUSSION PE enters cells by receptor-mediated endocytosis. Endocytosis is the beginning of the toxin pathway that results in the generation of an enzymatically active fragment, the translocation of this fragment to the cell cytosol and, ultimately, the inhibition of protein synthesis. Previously, we reported that PE binds to the heavy chain of LRP and most likely uses this receptor for endocytic uptake (14). LRP is a multiligand receptor whose function is the clearance and degradation of proteases and ligands related to lipid metabolism. Therefore, most of LRP ligands end up in the lysosome where they are degraded. For PE, some percentage of entering molecules some-how avoid this fate. Existence of an endoplasmic retention-like sequence at the C terminus of PE and results of mutational analysis of this sequence suggest that PE or its enzymatically active fragment is transported to the ER where it translocates to the cytosol.
Among the ligands that bind LRP, are RAP and the complex between tPA and PAI-1. To compare the fate of these ligands with PE, we constructed PE and DT hybrid proteins that have RAP or PAI-1 in place of the toxins' receptor-binding domains.
We found that DT-PAI-1 was able to inhibit the catalytic activity of tPA and form an SDS-resistant complex with this protease. This data suggest that the N terminus of PAI-1 is not essential for its interaction with tPA. Toxicity was clearly specific for the DT-PAI-1⅐tPA complex since the addition of DT-PAI-1 alone was at least 10-fold less active. However, compared with DT and PE, the DT-PAI-1 complexed with tPA exhibited a much lower level of toxicity for COS cells.
RAP-PE and DT-RAP were also less toxic than PE for mam- malian and insect cells. We showed that these proteins retained their binding activity for the heavy chain of LRP. Since the RAP cDNA is fused at the 5Ј end of the PE gene and 3Ј end of the DT gene, it appears that neither the N or C terminus of RAP are required for interaction with LRP.
Among various cell lines we did not find a correlation between PE and RAP-toxin sensitivities. Nevertheless, we were able to see such a correlation when the toxins were tested on isogenic cell lines. In particular, a PE-resistant line, 13-5-1, that lacked detectable LRP protein was less sensitive to RAPtoxins. A second mutant, 221-1, that expressed higher amounts of LRP than wild type cells and was supersensitive to PE by 3-5-fold, was also more sensitive to RAP-toxins. The lack of a good correlation between PE and RAP-toxin toxicities may be explained by results of several authors who have demonstrated the existence of additional receptors for RAP besides LRP. Indeed, a recent report has indicated that Chinese hamster ovary cells express the ␤ very low lipoprotein receptor, which in ligand blots, interacts with RAP but not PE (39). Thus in 13-5-1 cells, the toxicity of RAP-toxins, that is likely to be mediated by the ␤ very low density lipoprotein receptor, could be assessed. From the upper and middle panel of Fig. 7 it was clear that the toxicity of both PE and RAP-PE apparently relied heavily on the presence of LRP. However, this was not true for DT-RAP, because the 13-5-1 line was not greatly resistant to this hybrid toxin. Since PE translocates from the ER and DT from an acidic endosome, the intracellular transport requirements for these two toxins is likely to be quite different. One interpretation would suggest that 13-5-1 cells were not very resistant to DT-RAP because interactions with the ␤ very low density lipoprotein receptor were sufficient to mediate DT transport to an acidic endosome. RAP-PE was less active in 13-5-1 cells presumably because ␤ very low density lipoprotein receptormediated internalization did not facilitate transport to the ER.
To understand the reason for the lower toxicities of RAPtoxin chimeras, we checked the ability of RAP-PE to interact with LRP and to be cleaved by furin. RAP-PE appeared to interact with LRP with much higher affinity than the native toxin. At the same time substitution of the N-terminal portion of the receptor binding domain of PE by RAP did not dramatically change the ability of furin to cleave PE. It was therefore reasonable to suggest that either RAP-PE was not internalized by the cell or that after internalization, it followed a different pathway than that taken by native PE. To determine if RAP-PE was internalized to the endosomal compartment, we performed a co-incubation with adenovirus. We reasoned that if RAP-PE could reach the endosome, then adenovirus would enhance toxicity by releasing more toxin than was possible in the absence of virus. Toxicity was enhanced by 5-10-fold.
In conclusion, we speculate that a relatively low binding affinity allows native PE to dissociate from LRP and thus to avoid the fate of other LRP ligands, which is to be degraded in the lysosome. Our data showing increased cytotoxicity when RAP fragments were fused to PE instead of whole RAP seems to support this hypothesis. This finding may have wide reaching implications for the design of recombinant immunotoxins.