USEC ~ Y-DO COPY MONOCLONAL ANTIBODY ( MOAb ) THERAPY Use of anti-HIV immunotoxins as probes of the biology of HIV-infected cells

OBJECTIVE: Anti-human immunodeficiency virus (HIV) immunotoxins are potential treatments for HIV infection. but they may also be used as probes to study the relationship between HIV and the cell it infects. Data from the present study indicate the complexity of this relationship.

T IIE IIUMAN IMMUNODEFICIENCY VlRUS (HIV) ENVELOPE protein(s) are expressed on the surface of HIVinfected cells.These proteins consist of a precursor protein (gp 160).which is cleaved into extracellular (gpl20) and transmembrane (gp41) portions.Much of this protein is associated with budding virions.It is less well appreciated that a proportion of the envelope proteins is internalized from the cell surface.The processes of envelope protein synthesis, assembly and cellular trafficking are not yet fully understood.
The present study explored the potential of anti-HIV immunotoxins as therapeutic reagents (1-3).These immunotoxins consist of anti-gp 160 antibodies cou pied to ricin A chain (RAC).Other anti-HIV immunotoxins consist of the viral receptor, CD4, coupled to one of several different toxins (4 -6).These immunotoxins kill 1-1rvinfect.edcells in vitro.In vivo they would function by eliminating cells that are actively secreting viral proteins, thereby eliminating the nidus of infection.
Ricin immunotoxins kill target cells following internalization.Once inside the cell, the linkage between the antibody and RAC is cleaved.The free RAC functions in the cytosol to cleave the 28S ribosomal RNA and to halt.protein synthesis.This results in the death of the cell.Because immunotoxins must be internalized to function, they may be used as probes to study the cellular trafficking of the target antigen.The best evidence that exists demonstrating that the HIV envelope protein is internalized is that even monovalent HIV-specific immunotoxins function to kill infected cells.The only way for monovalent immunotoxins to enter and kill the cell is for the ligand to which the immunotoxin is attached to be internalized, most likely as part of its cellular trafficking.In the present study, a panel of anti-HIV immunotoxins was used to study the cellular trafficking of the HIV envelope proteins.A number of factors was found to be able to influence the internalization and intracellular routing of these immunotoxins.

MATERIALS AND METHODS
Antibodies, immunotoxins, mv strains and cell lines: The antibodies and immunotoxins used in this study have been described in detail elsewhere (2).The antibodies are listed in Table 1.Immunotoxins were constructed by coupling the indicated antibody to purified RAC using the heterobifunctional cross-linker N-succinimidyl -3-(2-pyridyldithio) propionate.This cross-linker contains a disulfide bond that. is broken upon reaching the reducing conditions found within the cell.The immunotoxin CD4-PE40 , a chimeric protein consisting of the 24A two amino-terminal domains ofCD4 fused to a truncated form of pseudomonas exot.oxinlacking the binding domain (4), was obtained from Upjohn Laboratories (Michigan).Persistently HIV-infected cells were established using the molecular clone NL4-3 to infect the H9 T cell line (2, 7).Acute infections were established in A3.0l cells with a variety of strains of HIV and then adding fresh uninfected cells t.o the infected cell culture weekly at a ratio of 10 uninfected cells t.o one infected (2).Functional assays of immunotoxin action: The effects of the immunotoxins on infected cells were assayed in two ways: by cytotoxicity and by inhibilion of production of infectious HIV.Details of these assays are described elsewhere (2) .In persistently infected cells, where nearly all the cells express viral antigens on the cell surface, cytotoxicity could be assayed as inhibition of protein synthesis.Cells were incubated with t.he in1munotoxin for 48 h, pulsed with 35 S-methionine and harvested 16 h later.A focal immunoassay was used to measure the production of infectious HIV by both persistently and acutely infected cells.Cells were incubated with the immunotoxin for 24 to 48 h, washed and transferred t.o monolayers ofHeLa-co4+ cells (8,9).One day later the monolayers were washed, then incubated for an additional two days.The monolayers were then fixed with ethanol and stained with human anti -HIV serum and peroxidase conjugated goat anti-human immunoglobulin.Foci were visualized with the colorimetric substrate aminoethylcarbazol.Cell surface antigen expression: Flow cytometry was used to study the expression of 111V antigens on the cell surface (2).Cells were p laced in phosphate buffered saline containing 1.0% bovine serum albumin and 0.01 % sodium azide.Cells were incubated in the test antibody at 10 µg/mL for 1 h, washed and stained with fluorescein conjugated goat anti-mouse or human immunoglobulin for 1 hand washed.Five thousand cells   Western blots: Cells were lysed and run on SOS-PAGE, blotted onto nitrocellulose, then incubated wit.h affmity purified human anti-gpl60 antibodies.Immunoreactive bands were detected with 125 1-prot.einA (10).

Antibody and virus effects on immunotoxin action:
We have clearly demonstrated in other publications that many , but not all, anti-gpl60 antibodies can function as irnmunot.oxinswhen coupled lo RAC (1-3).IL is clear that t.he antibody must.bind to an epitope that is exposed on t.he surface of the infected cell.If there is no cell surface binding, the immunotoxin cannot function.However, we have found that.the efficacy of the immunotoxin is not.always related to the degree of cell surface binding.This is demonstrated in Table 2. Antibody 41. l bound to the cell surface only marginally, yet functioned as well as an irnrnunot.oxinas antibody 924, which was highly expressed at the cell surface.Conversely, antibody 110. 1, which bound to t.he cells moderately well, was ineffective as an immunotoxin.We termed this phenomenon the antibody effect.and have demonstrated that.this was not a function of the avidity of the antibody.This has a lso been observed with immunoloxins directed against other target antigens and may be a function of the proximity of the epitope to the cell surface (11).If proximity is an important.determinant oft.he antibody effect., then these immunoloxins may be used to map t.he topology of gpl20 as expressed on the surface of infected cells (and presumably on the surface of virions).The data would !.hen indicate that the V3 loop lies close to the cell membrane and would explain the surprisingly good efficacy of anli-gp41 immunoloxins.
We have also found that.different st.rains of virus have different.degrees of s u sceptibility lo an immunoloxin.Paradoxically, this effect is most.pronounced in virus strains that produce inlense cell surface fluorescence and are minimally susceptible to the immunotoxin.As shown in Table 3, cells infected with strain 22 stained brightly, yet were resistant lo the effect.oft.he immunotoxin.This effect, which we tem1 the virus effect, most likely represents t.he balance between virus secretion and antigen internalization.Cells that s lain very brightly may be producing large quantities of budding virus.This virus then binds the immunoloxin and prevents it from reaching the smaller quantity of antigen that is internalized.We h ave observed this effect with murine retroviruses as well.Th e suggest.ion that the ratio of secreted virus to internalized antigen can vary with different strains of HIV within t.he same cell type indicates the complexity of t.he virus/host cell relationship.

Immunotoxin-resistant mv variants:
We have used immunotoxins to select.persistenUy infected cell lines for variants Lhat can resist irnrnunot.oxinact.ion.We have previously shown that two different types of variants can be selected: those failing to express any viral proteins and those in which the expression of the envelope protein is specifically altered (12).In our initial report, the variants wiU1 altered envelope expression had a mutation in the envelope gene that resulted in truncation of gp4 l, which in turn altered the processing of envelope protein the intracellular routing of the virions.We have now performed additional selections as summarized in Table 4. Cell lines lOE and 4F were described in the previous publication and represent examples of each of the two classes of variants.Cell line C9 does not express HIV proteins because it has lost the integrated provirus from its genome.G4 represents a class of variants that have a completely normal envelope gene, but do not express the protein at the cell surface.As indicated in the Western blot shown in Figure 1, G4 produces and processes full length gpl20 and gp4 l .We have sequenced the entire gene, and it is identical lo the gene in the parental cells.Yet, the envelope protein is not expressed al the cell surface.Most likely there is a mutation in the cell line carrying this virus, and this affects the processing of the protein.Yel another lype of variant is exemplified by clone 4. As opposed to the other in1munotoxin resistant variants that do not produce infectious virus, these cells produce high levels of infectious virus.The alteration in the expression of the HIV envelope in clone 4 that allows for immunotoxin resistance, yet retains infectious properties , has not been fully elucidated.By using irnrnunotoxins to select for variant HIV-infected cells, we have been able lo produce a series of alterations in either the virus or the cell carrying the virus.These have been extremely informative in defining the processing of the HIV-envelope protein (12).Effect of soluble CD4 on anti-gp41 immunotoxins: We have recently reported thal incubation of HIV-infected cells with soluble CD4 greaUy enhances the efficacy of anti-gp41, but not anti-gpl20, immunotoxins (3).This effect is due to two different processes.The first is the phenomenon of CD4 stripping of gpl20 from gp41 (13 , 14), which results in an increase in exposure of gp41 epilopes on the cell surface and a decrease in gp120.
The second effect is that incubation with soluble CD4 enhances the rates of internalization of both gpl20 and gp41.These two effects are additive for gp41, resulting in enhanced efficacy of anli-gp41 in1ffiunotoxins.For gpl20, the effects cancel each other out, resulting in no real change in efficacy of anli-gpl20 in1ffiunotoxins.These observations have obvious therapeutic implications.They also demonstrate that binding of the envelope protein to its natural ligand can influence the cellular trafficking of the envelope protein.

Effects of lysosomal inhibitors on immunotox in activity:
Once an immunoloxin is internalized, the path it follows determines its ability lo kill the target cell (15) .Intracellular routing may return it lo the cell surface, send it in a retrograde fashion through the protein secretory pathway, or direct il lo the lysosomes where it will be degraded (16).To function, the toxic moiety must eventually be present within the cylosol.Characteristics of both the in1ffiunoloxin and the target antigen determine what proportion of the immunotoxin will follow each pathway.Drugs that inhibit lysosomal deg-

USE O LY• DO OT COPY
Anti-HIV immunotoxins radation of the immunoloxin can be used to determine whether the lysosomal pathway plays an import.antrole in the routing of a particular inununoloxin.They may also be used to enhance the pharmacological effects of the inununotoxin.We have tested U1e effects of three different lysosomal inh ibitors on anti-HN immunotoxins.Ammonium chloride alkalinizes the lysosome, inhibiting enzyme function.Chloroquine stabilizes the lysosomal membrane, preventing transfer across it.Monensin reduces ion flux across membranes and inhibits transport into lysosomes, as well as interfering with intracellular trafficking of endosomes and secretory vesicles.As shown in Figure 2, all three agents enhanced the efficacy of 924-RAC, monensin and chloroquine enhanced 41.4-RAC, but only monensin enhanced CD4-PE40.These data demonstrate that the lysosomal pathway plays an important role in the intracellular routing of antibody-ricin immunotoxins, but probably not of CD4-PE40.

CONCLUSIONS
Because anti-HIV immunotoxins must both bind to the surface of HIV-infected cells and be internalized function, they may be used as probes of the cellular trafficking of lhe target antigens.Thal these immunotoxins function at all is Lhe strongest indication that there is a recirculating pool of envelope protein that re-enters the cell after being displayed on the cell surface.A number of factors, including Lhe strain of 111v and Lhe presence of a soluble fom1 of the 111v-receptor CD4, can influence the degree of internalization.Once internalized.the lysosomal pathway is an important route in the processing of antibody immunoloxins bound to the envelope protein.Using lhe immunotoxins to select variants, we have identified mutations within both the virus and the cell lhal can affect the processing of lhe envelope protein and alter the secretion of the virus.All of these data indicate lhe complexity of the relationship between 111v and U1e cell it inhabits.
The data described in this article also have important therapeutic implications.The ability to enhance the efficacy of antibody immunoloxins with soluble CD4 and with lysosomal inhibitors, wilhoul a consequent increase in nonspecific toxicity.should augment any therapeutic utility of the immunotoxins.Initial clinical trials with anti-HIV immunotoxins are in progress.

Figure 1 )
Figure 1) Weste rn blot of immunotoxin resistant variant cells: cell lysates w ere run on SDS-PAGE, blotted onto nitrocellulose and d etected with anti-gpl 60 antibodies.H9 cells w ere uninf ected , H9/NLA-3 w ere the parental cells of the variants.and G4 and B9 the immw10toxin-resistant variant cells.Bands corresponding to the full length gp 120 and gp41 are s een in both the parent and the variants.B9 also contains an unidentif1ed band migrating between the two, which may repres e nt a gp120 or gp 160 d egradation product

resistant variants selected from persistently HIV-infected cell lines
+were analyzed for fluorescent.inlensity on a Beet.onDickinson FACStar (California).