Jay W. Chaplin

Reprinted with permission from Churchill Livingstone Journals. Medical Hypotheses, Feb. 1999, Vol. 52, No. 2, pp. 133-146. More information about Medical Hypotheses is available at the Medical Hypotheses web site.

Contents

Summary

Introduction

Response of the Human Immune System to HIV

The Roles of HIV and gp120 During Infection

HIV and gp120

HIV, B Cells and Antibodies

Qualitative and Quantitative Effects of HIV Infection on T₄ Cells

A Possible Cause, a Probable Mechanism for HIV Pathogenisis

Anergy

Apoptosis

Clonal De-Selection and Immunotoxin Design

References

Acknowledgements

Summary

Current models of HIV's role in AIDS pathogenesis are inadequate and inconsistent with the literature. This article reviews a wide range of AIDS research and proposes the first model of HIV pathogenesis that is entirely consistent with the literature. This model is based on antibody-gp120 complex crosslinking of CD4 on the T₄ cells. Previously unexplained observations embraced by this model include early qualitative defects in the immune system cells. Previously unexplained observations embraced by this model include early qualitative defects in the immune system, changes in cytokine expression, "bystander" T₄ cell death, and the apparent discrepancy between the low rate of T₄ cell infection and the near-complete elimination of T₄ cells in AIDS.

A new class of drugs based on this model is detailed. These drugs should disrupt the pathway leading to AIDS and leave an HIV infection indefinitely asymptomatic. These drugs are designed to be readily modified, so the treatment is immune to HIV's notorious mutation-based drug resistance.

Introduction

In the sixteen years since AIDS was first recognized, we have learned a great deal about the epidemiology and consequences of this syndrome. The vast majority of the scientific community has embraced HIV infection as the cause of AIDS, and the development of anti-HIV drugs has brought a measure of relief to many patients. However, the discovery of a clear and accessible etiologic cause is not the same as clarifying the biochemical pathway to disease. The current model of "direct infection", which holds that the immune system is destroyed by the process of HIV infecting T₄ cells and killing them during viral replication, has significant inconsistencies. Beyond the surprisingly low frequency of HIV infected cells that supposedly result in a near total depletion of T₄ cells, there are other inadequacies in the direct infection model. It is the aim of this paper to clarify these additional inconsistencies, to present another model by which we believe HIV causes AIDS, and to propose a new class of therapeutic drugs based on that model.

Keeping these goals and the great diversity of opinion about biochemical cause in mind, it may be useful to state some of the indisputable relationships and facts - the phenomenology of AIDS - which any acceptable theory must account for.

Response of the Human Immune System to HIV

During the primary HIV infection, large quantities of the virus are produced (equivalent to that of full term AIDS) and the number of helper T (T₄) cells declines precipitously. After a short period (8-40 days) a vigorous HIV-specific cytotoxic T (T₈) cell response develops and there is a concomitant drop in viral titer (1-3). Indeed, the earliest manifestation of an HIV infection, the T₄ / T₈ ratio drop, begins with an elevation of T₈ cell number rather than a loss of T₄ cells (4). There is no reason to believe that significant neutralizing (virus inhibiting) antibody titers develop during this time, and it seems that the first antibodies to be produced are highly strain-specific and quickly lose efficacy as the initial HIV strain diverges into a host of related quasi species (3, 5, 6). These viral mutants that temporarily escape immune surveillance will continue to appear -- the mutation rate of HIV, like the rest of its lentivirus family, is over a thousand fold greater than that of the viruses responsible for the common cold (7). This high mutation rate is due to the lack of exonuclease proofreading and the lack of copy fidelity in the reverse transcription following infection (8). Antibodies to the few constant (non-mutating) and broadly recognized epitopes of HIV's surface glycoproteins seem to occur much later (9).

During the initial infection, the interleukin 2 (IL-2) production by the T₄ cells begins to drop (10). Over the next two years, the IL-2 production drops by half, and continues to decrease (10, 11), while levels of interleukins 4 and -10 increase (12) resulting in polyclonal B cell activation and hypergammaglobulinemia (overproduction of antibodies) with both HIV-specific and random antibody expression (13). This shift from type 1 (IL-2 & IFN- ) to type 2 (IL-4, -6, and -10) cytokine hormones, and consequently from T₈ and natural killer cell to B cell responses (14), is poorly understood but typical of progressors in AIDS (10, 11, 15, 16). This hormone imbalance reinforces itself, as IL-10 has a regulatory feedback loop that further inhibits the expression and action of IL-2 (17). Nonprogressors (long term asymptomatic HIV patients) seem to retain some measure of IL-2 production (18, 19).

As the immune response changes from cell-mediated to antibody-mediated there is a continual and profound loss of T₄ cell function, and to a lesser extent, a loss of T₄ cell numbers (20). Curiously, the resurgence of HIV in the peripheral blood seems to correlate with neither a particular time span, nor the number of T₄ cells infected or depleted, but with the loss of T₈ cell efficacy (21). The continually progressing impairment of B cell responses to new antigens and total loss of T₈ cell function leads to immune collapse and the opportunistic infections characteristic of AIDS. While the symptoms we recognize as AIDS are clearly related to the death of T₄ cells, they cannot be caused solely by that T₄ cell death. In order for T₄ cell death to have a causative role in immunosuppression it must occur prior to symptoms, and this relationship is not supported by the literature that continually shows broad immunologic defects appearing prior to T₄ cell loss (22). Some other mechanism must be operating, one that accounts for both an early immunologic defect and for late-stage T₄ cell death.

The Roles of HIV and gp120 During Infection

HIV and gp120

One of the primary differences between the various strains of HIV-1 (in the USA) and HIV-2 (primarily African) is the particular form of the surface glycoprotein gp120 that they carry. This gp120 protein forms the outer coat of the virus during budding from the cell. It is extruded through the membrane of infected cells, attached only by its weak interaction with the transmembrane protein gp41 (23). Since the membrane of HIV is formed from the membrane of an infected cell, the composition and organization of these surfaces will be similar (24). Because gp120 is the only viral protein significantly exposed to blood and lymph it plays a singularly important role, both as the principal neutralization site for antibodies and as the initiator of viral entry and infection. HIV binds to the cells it selectively infects by interaction with its gp120, either to the CD4 protein on T₄ cells and macrophages (25) or to the galactosyl ceramide lipids of colonic epithelial cells, sperm cells, neurons, and related brain cells (26-29). While the cells infectable through galactosyl ceramide are all equally susceptible if they have the same concentration of that lipid, the CD4 pathway is different. The particular segment of gp120 that determines cell infectivity (tropism) for T₄ cells or macrophages happens to reside in one of the most mutation-prone variable segments of gp120, the V3 loop, which also elicits the most antibody response (30, 31). The continual mutations in this V3 loop region seem to drive the transition from the initially infective macrophage-tropic HIV strains to the T₄ cell-tropic strains that usually appear near the onset of AIDS.

Complicating matters more is the fact that the connection between gp120 and its membrane anchor partner, gp41, is so tenuous. Of the ‚288 gp120 proteins on the surface of a newly created HIV particle, about half will be lost before the virus either infects a cell or is eliminated by the immune system (32). Little is known about the quantity or behavior of gp120 on the surface of infected cells. However, there must be enough gp120 to permit the creation of virus by budding, and this gp120 will also be shed (33, 34). The shedding of gp120 both hurts and helps the virus. A loss of gp120 is a loss of the virus' targeting / fusion mechanism for infecting T₄ cells, and also a loss of the primary target for antibody-dependant lysis. While shedding gp120 reduces the infectivity of the virus, it also "distracts" antibodies from the virus by adding separate soluble antibody targets to the bloodstream. Unfortunately, the binding of antibody or CD4 to gp120 significantly enhances the likelihood that the gp120 will disassociate (or "shed") and float freely through the bloodstream, further protecting the virus or infected cell from lysis (35, 36). This enhanced shedding phenomenon is likely responsible for the low infection rate of HIV, as gp120 loss significantly reduces the virus' ability to infect a target cell. The same evasion of antibody lysis also enhances HIV's extraordinary resistance to antibody neutralization; if each antibody removes gp120 rather than directly "killing" the virus more antibody will be required to eliminate the virus. As each antibody is bivalent and all gp120's on an HIV particle must be blocked or shed before the virus can be assumed noninfectious, it must be bound by a minimum of 144 antibodies. Without efficient lysis, complete virus neutralization requires one antibody for every pair of gp120s.

A further complication is that the propensity of HIV and infected cells to shed gp120 is not constant. In addition to determining the viral tropism with the V3 loop, each strain also varies in the attachment strength of gp120 to gp41. Therefore, each strain varies in its infectivity and susceptibility to antibody neutralization. Each of these characteristics, cell tropism and gp120 shed rate, can be changed independently through cassette mutagenesis techniques in the lab, but they always seem to occur together during evolution of HIV strains in the body (31). It appears that macrophage tropic strains of HIV have a lower gp120 shed rate than T-cell tropic strains, and consequently are more infectious, easier to neutralize, and release less gp120 antigen to the bloodstream (31, 37). Conversely, the T₄ cell tropic strains that develop during long term HIV infection are less infectious, exceptionally difficult to neutralize, and tend to release large quantities of gp120 into the bloodstream -- mostly in the form of antibody complexes (38). Even in the absence of free virus, as during protease inhibitor / anti-retroviral combination therapy, a smaller but significant level of gp120 is maintained in the bloodstream by the infected cells. This is unavoidable, as the gp120 protease processing is done via essential host cell membrane proteases such as furin (39), rather than by the HIV protease targeted by protease inhibitors.

The gp120 released from infected cells and virus is quickly cleared from the bloodstream, as it must be to maintain a steady plasma concentration in the face of constant rapid production and shedding. While some gp120 is deposited in antibody complexes throughout the kidneys and other tissues, this is a minor pathway for gp120 removal until the development of AIDS and/or T₄ cell tropic strains of HIV (40). Through binding to B-cell membrane immunoglobulin receptors and to the CD4 protein of T₄ cells, the free gp120 is removed from circulation and coated onto the B and T cell surfaces. Within a few hours after binding to the cells, the gp120 is internalized (41), either to be degraded via the major histocompatibility complex II (MHCII) lysosomal pathway of B cells or through a standard lysosomal proteolysis and ejection in T₄ cells. This processing manages to keep the total concentration of gp120 in a person with full blown AIDS at or below 40 nM. This clearance mechanism adds two more layers of complexity: rapid degradation of gp120 with a down-regulation of CD4, and a shift in cytokine hormone expression from the T₄ cells.

While the rate of gp120 clearance by B cells has not been documented, it has been well studied in T₄ cells. There the surface-bound gp120 is maximal after one hour and decreases by ‚35% over the next 18 hours. Since each gp120 is internalized while bound to its CD4 receptor, there is also a 35% decrease in surface CD4. Immunologic staining has shown that both the persistently absent CD4 and the internalized gp120 can be found, still in a bound state, within the lysosomal compartment. In the course of constant extracellular exposure to gp120, the same ‚35% of CD4 receptors can be expected to remain internally tied up through their association with the lysosomal burden of gp120 (41). With HIV infection and its continual internal overload of gp120, even more CD4 is retained within the cell, leading to a total loss of CD4 cell surface expression (42, 43). Since CD4 mediated signals are critical for proper T₄ cell activation, and the regulation of the rest of the immune system through them, the loss of accessible CD4 will certainly affect cell signaling and function throughout the immune system.

By internalizing and clearing gp120, the T₄ cells expose themselves to another immune system defect in addition to the CD4 deficit. When gp120 is bound to CD4 and internalized, it causes a dramatic decrease in the production of IL-2 and a reduction in expression of the IL-2 receptor (44). As IL-2 is a broadly active and powerful cytokine hormone, these events dramatically change the regulation of the entire immune system. When IL-2 production and response drop, the results are defects in T₄ cell proliferation, natural killer cell action (including antibody-dependant cell lysis), and T₈ cell function (45). These defects, in and of themselves, can be corrected in early stages through restorative dosing with IL-2, though such dosing in HIV patients with T₄ counts of less than 400 does not seem to be effective (46). As the T₈ and natural killer cell portion of the immune system weakens, the reciprocally regulated antibody-producing B cells over-proliferate and differentiate (47). These B cells will continue to over-respond due to the presence of a persistent supply of the gp120 antigen that drives their responses.

HIV, B Cells and Antibodies

The effects of an HIV infection on B cells are perplexing. Although HIV does not infect B cells, the presence of HIV slowly but drastically increases B cell activation and antibody secretion. In the normal primary response to a soluble antigen it takes 5-7 days to activate 0.001 to 0.01% of the B cell population and build a measurable antibody response. In the case of B cells responding to gp120 from HIV's coat, it requires 11-14 days to achieve the same level of antibody production (45). Yet HIV post-infection activation percentages for B cells have been measured in the range of 6%, a response of perhaps 6,000 times the norm for other common antigens (48). The rate of B cell response is slow but the quantity of response is large.

In HIV vaccine experiments there has been great frustration over the poor immune response and antibody production obtained from the injection of purified gp120. This slow and low antibody response is probably due to the rapid removal of gp120 by the large quantities of T₄ cells before the 6% of reactive B cells have a chance to encounter the gp120 and begin their proliferation and differentiation process. This rapid clearance also makes it difficult to assess the safety of gp120 subunit vaccines, as the safe infrequent injection exposure with quick clearance is difficult to extrapolate to the effect of a live attenuated or recombinant virus vaccine that would produce significant quantities of gp120 continuously and indefinitely.

One of the most interesting features of the antibody response to gp120 is that it is not completely specific for gp120 and can lead to polyclonal activation (non-specific B cell activation), followed by random antibody expression (49). This, plus the systemic overexpression of IL-4 that occurs with HIV infection, a condition that favors antibody overproduction and polyclonal expansion of B cells (50), would seem to set the stage for B cell mediated autoimmune diseases. Expectedly, long term HIV infection carries a greatly enhanced risk of autoimmune disease, with the production of antibodies to nuclear factors, fas, CD4, MHC1 and several others (51-53). Indeed, the diagnosis of HIV / AIDS has often been misdirected by symptoms and test results indicative of SLE (systemic lupus erythematosus) (54).

While the initial production of antibody to gp120 is slow, the long-term response is vigorous, leading to reported highs of 14-fold increases in IgM and 34-fold increases in IgG antibody classes and persistent hypergammaglobulinemia (55). For the fraction of antibodies directed against gp120, the standard process of somatic mutation and titer enhancement continues as long as there is adequate T cell help (9, 56). However, this process of antibody refinement seems to happen too slowly to keep pace with the mutation rate for variable segments of HIV's proteins. As a result, the virus mutates out from under the control of the immune system.

One reason the antibody response to gp120 is so ineffective is that most antibody production is against the immunodominant V3 loop (57) -- the most variable region in gp120. In several experiments it was determined that antibodies to the variable regions had poor neutralization efficiency and, due to their poor efficacy in recognizing mutant viral strains and slow generation rate, tended to target HIV strains that were not infectious (9, 58, 59). This, combined with the requirement for large quantities of antibody to insure virus neutralization, limits the utility of an antibody response. While V3 antibodies are the first to be produced and achieve a high titer, they become ineffective as the HIV strains diverge with a rapid rate of mutation, and the fraction of virus neutralizable by that particular antibody diminishes.

While the small percentage of antibodies directed at gp120's constant (slow or non-mutating) regions seem more promising, they present other problems. In addition to the ability of some antibodies to enhance the likelihood of HIV infection, the lysis mechanism of antibody dependant cellular cytotoxicity (ADCC) seems to be impaired. In experiments using both natural killer effector cells from HIV+ patients and effector cells from healthy volunteers, the patient cells were found to be defective in mediating ADCC. Apparently this defect arises from a lack of killing factor release for destruction of the targeted cell, and is likely due to maturation of the natural killer cells in the absence of IL-2. The efficacy of natural killer cell mediated ADCC is partially restored by adding exogenous IL-2 (60, 61).

While the inefficacy of ADCC is a hindrance to control of HIV, it may be a benefit on the whole as anti-HIV antibodies mediate undesirable effects. Studies on ADCC in people with AIDS have found that uninfected T₄ cells were better targets for ADCC than free virus or infected cells, apparently due to the greater stability of gp120's binding to the CD4 receptors on T₄ cells than to gp120's viral partner gp41 (16). If the gp120 associated with CD4 on uninfected T₄ cells can serve as an antibody target, then the immune system participates in its own destruction. Indeed, cytofluorograpic methods have shown that significantly more gp120 is associated with the surface of uninfected T₄ cells incubated in gp120 than with the surface of infected cells producing gp120, and that this correlates with a preferential lysis of the uninfected cells (62, 63). Additionally, further experiments showed that 100 fold less antibody was required to mediate ADCC of uninfected cells than infected cells (64), and that the concentration of gp120 in an AIDS patient was more than 80 times higher than the concentration needed to mediate maximal ADCC (65). Clearly, the uninfected cells will be eliminated by ADCC faster than the infected cells because of the higher density of gp120 antibody target and greater susceptibility to antibody-dependant lysis. So the ADCC clearance mechanism favors the elimination of healthy uninfected cells over those that are infected. It attacks the immune system more often than the virus, thus ADCC is non-functional and best left crippled.

Qualitative and Quantitative Effects of HIV Infection on T₄ Cells

The current model of AIDS is that HIV directly kills T₄ cells through infection. This hypothesis seems congruent with the emergence of T₄ tropic strains of the virus at the time of the development of AIDS. As the T cell tropic strains would be expected to cause greater infection and death of T₄ cells than the macrophage tropic strains responsible for initial infection, this analysis seems reasonable. There are, however, T₄ cell effects that this model cannot account for. These include: infection / death discrepancies, early qualitative defects of T₄ cells, and bystander T₄ cell death.

One of the most puzzling questions in HIV pathogenesis is how an infection limited to small fractions of host cells can cause such severe consequences to overall health. There is much debate over the actual number of cells infected, as each method for determining HIV+ cell numbers seems to present significantly different results. However, all results appear low for the quantity of cell death produced. To date, the highest total count of HIV+ T₄ cells reported in peripheral blood from an asymptomatic patient has been one in ten thousand (66), while the number of these cells during fulminant viremia and AIDS has been only as high as one in two hundred (67). The number of HIV+ T₄ cells reported in lymph node tissue is consistently higher, and peaks at one in one hundred (66). It remains difficult to imagine that an infection starting with a 10-4 frequency and detected in, at most, one percent of the T₄ population could cause an 80% or greater elimination of these same cells (55). When this is viewed in light of the quick clearance rate for free virus (68), the prodigious replacement rate of T₄ cells (on the order of the rate of virus production in full blown AIDS) (68), and the fact that only one out of one hundred to four hundred of all HIV+ cells is productively (versus latently) infected (67, 69), the direct infection model becomes more tenuous.

While these figures do not rule out the direct infection model, which could fit these data on a kinetic basis of extremely rapid removal of infected cells, the data do show a need for caution in interpretation. Also problematic are the data from primary infections where 1% of the peripheral blood T₄ cells are HIV+ (70). This is higher than in full blown AIDS, and yet the immune system recovers to provide eight to ten years of clinical remission. This discrepancy between early and late effects of equivalent amounts of HIV infection seems highly inconsistent. It has been assumed that the later development of AIDS is due to the production of escape variants of the virus, which are temporarily unblockable by the immune system. This interpretation is inconsistent as the rate of mutation and escape variant production is determined by the rate of infection and reverse transcription, so both rates would be as high during the primary infection as they are in full term AIDS. Thus, the presence of escape variants does not adequately explain the difference in symptoms between acute and chronic HIV infections. Furthermore, in infections with other members of the lentivirus family the production of escape variants does not correlate with a disease state (71). The development of strain differences is also incapable of resolving this issue, as there are well-documented cases of long term non-progressors carrying highly cytopathic strains (72) as well as studies clearly showing that the rate of T₄ loss is not directly determined by the viral replication rate (68). These significant inconsistencies remain.

In addition to these quantitative problems with the direct infection model, there exists the inescapable problem of early T₄ cell qualitative defects (20, 73). In the asymptomatic phase of the HIV infection there is a continuous and progressive development of T₄ cell proliferative defects and anergy (loss of activation potential) while the T₄ cell numbers are still well within normal bounds (74-76). Indeed, even when identical numbers of purified T₄ cells are tested, a selective defect in antigen recognition was noted; even the living T₄ cells do not function properly during an HIV infection (74, 76). This lack of function on the part of T₄ cells is difficult to resolve with any model based solely on death of those cells; the direct infection model recognizes only uninfected functional cells, killed cells, and a minority of infected cells that might have reduced functionality. While the T₄ cells can be made to function in the presence of gp120 in vitro by restorative dosing with IL-2, their inability to recognize foreign compounds in vivo is not completely reversible, and the IL-2 treatment is surprisingly dependant on the number of remaining T₄ cells for its efficacy (46, 54).

The largest barrier to the credibility of the direct infection model, however, is the "bystander effect" in HIV-induced cell death. In studies of apoptosis (programmed cell death, PCD) in the lymph nodes during HIV infection, where a correlation between the infection of a cell and the cell's death was expected, it was consistently found that the cells that were dying were those infectable but not yet infected by HIV (77, 78). While the cells infected by HIV do indeed die at an accelerated rate, this is of minor significance compared to their ability to induce PCD quickly and efficiently in large numbers of healthy nearby cells. These results conflict with the direct infection model of HIV-induced cell death.

The previously mentioned IL-2 deficiency, excess of IL-4 and -10, and the inability of restorative IL-2 dosing to correct for IL-2 controlled immune defects is a significant hurdle for any model of AIDS. While the binding and internalization of gp120 certainly contributes to the IL-2 related defects and does occur extensively, with an average of 8,000 to 10,000 molecules of gp120 bound to CD4 per T₄ cell at any given time (79, 80), these effects are totally reversible with exogenous IL-2 (44, 81). The presence of normal IL-2 levels allows T₄ cells completely coated with gp120 to function normally, yet it does not restore T₄ cell function in people with AIDS. As to the enhancement of IL-4 and -10, few theories have been set forward.

Further confusion comes from the fact that T₄ cell death manifests differently in pure T cell culture compared to peripheral blood cells (PBMC). It is surprising, and difficult to understand from the view of a direct infection model, why the presence of other cell types would affect the extent and manifestation of cell death, yet they do. In the clearest experiment using these methods it was found that when the majority of a pure T₄ cell population was infected, the infected T₄ cells died. However, in a PBMC culture with the same level of HIV infection, both infected and non-infected T₄ cells were dying (82).

Any effective model of HIV pathogenesis must have a mechanism to support bystander cell death, a 10+ to one kill per infection rate for T₄ cells, the IL-2 irreversibility of immune system defects, and the anergy encountered in living T₄ cells from people with HIV infections.

A Possible Cause, a Probable Mechanism for HIV Pathogenisis

Despite such severe problems with the direct infection model, there has been little focus in the research community on alternate models of HIV-induced T₄ cell death and immune system dysfunction. We suggest that the antibodies produced against gp120 can contribute to the defects and death of T₄ cells. In addition to the possibility of direct killing of uninfected cells through ADCC, it appears that there are at least two other ways that the binding of antibody / gp120 complexes can damage the T₄ cells.

While direct effects such as ADCC demand only that the antibody be bound to the target, each antibody has the capability to bind to two identical target molecules. This dual binding both increases the strength of association and physically forces the two target molecules into persistent close contact, called "crosslinking". If the target molecule is physiologically active, e.g., a receptor or signaling complex, then this crosslinking may have serious secondary effects. Immune complexes of gp120 and antibodies (gp120 IC) bind to the surfaces of T₄ cells through their association with CD4, in a trimolecular sandwich: CD4, gp120, anti-gp120 antibody (83). As CD4 is known to transduce powerful signals to the T₄ cell and to vary the type of signal delivered depending on the type and order of stimulation relative to other receptors (84-88), this crosslinking of CD4 could be significant (89). The signaling of CD4, normally occurring when it binds to antigen:MHCII complexes along with the T cell receptor, is required for the proper activation of naive and memory T₄ cells. While CD4 crosslinking was originally thought to hinder binding to MHCII, we now know that it can change the signal delivered without blocking MHCII contact (85), and that gp120 can bind without blocking MHCII (90).

There is a clear example of CD4 crosslinking in the literature; anti-CD4 antibodies have been used during transplantation and autoimmune disease therapy to induce temporary immune system collapse and persistent tolerance to particular antigens (91). Given the importance of their action on the immune system, these anti-CD4 antibodies have been thoroughly investigated. They produce two defects in T₄ cells, the same two, anergy and apoptosis, as gp120 IC. (47, 84-86, 92-95). In addition to providing a parallel case for investigation, these anti-CD4 studies point out a significant and infrequently recognized flaw in many experiments involving T₄ cells: Positive selection of T₄ cells by anti-CD4 binding significantly and persistently modifies their behavior and inherently biases experimental results (87). Only those experiments with negative cell selection can be assumed to give physiologically relevant data. All others must be scrutinized. Such factors require diligent attention and must be assessed separately for each experimental protocol. Therefore, caution is required in interpreting the AIDS/HIV literature.

Anergy

Anergy, the inability to respond to appropriate stimulation, manifests itself in several forms: The cell may be totally unresponsive; antigen recognition may be impaired either globally or specifically; activation induced cell proliferation may be blocked; or the patterns and levels of cytokines may be modified -- with cytokine effects leading to a shift in responding cell types during an immune response. Any combination would fall under the overall classification of anergy, and all can be caused by either anti-CD4 treatment or gp120 IC crosslinking.

While the lack of an immune response against new antigens is a hallmark of HIV infections, often occurring before AIDS has developed, there has been little attempt to explain this process. The direct infection model has no explanation for this, but antibody crosslinking does. Based on the assumption that gp120 IC crosslinking of CD4 could cause the same inactivity of T₄ cells as anti-CD4 antibodies, a set of experiments comparing immune complex dose response to calcium ion mobilization was carried out (96). The intracellular mobilization of Ca++ was flourescently monitored as a direct marker of T₄ cell activation, while varying concentrations of gp120 and anti-gp120 antibody were added to determine their dose response and effect on T₄ cell activation. The T₄ cells clearly failed to mobilize calcium when pretreated with both gp120 and antibody from HIV+ individuals, but not when treated with either gp120 or antibody alone. Repeated tests showed that antibodies from 14 of 14 HIV+ individuals mediated this immunosuppressive action in concert with gp120, while control antibodies from patients infected with cytomegalovirus, HTLV-1, rubella, and hepatitis consistently failed to do so. In anti-CD4 antibody crosslinking, this form of induced anergy affects both naive and active / primed T₄ cells (95). These results demonstrate that gp120 immune complexes can block T₄ cell activation, and they provide a mechanism for the unexplained qualitative defects of uninfected T₄ cells in AIDS. As HIV infected T₄ cells do not express CD4 on their surfaces, they are immune to this effect, only uninfected T₄ cells exhibit this effect.

It is puzzling that AIDS research focuses on the elimination of T₄ cells, since elimination is only one regulator of their numbers in the body. The normal process of an immune system response includes proliferation, or "clonal expansion", of recently activated T₄ cells, increasing the number of T₄ cells specific for the invasive antigen at the site of infection. This proliferation is inhibited in AIDS (83, 97). Thus the drop in T₄ cell numbers is caused by both a faster elimination and a slower production. This same type of effect is seen with anti-CD4 antibody dosing: even antibodies that do not lyse target T₄ cells prevent the proliferation of T₄ cells on immune system challenge with a mismatched transplant (85, 94). In CD4 binding experiments it was found that monomeric binding of CD4 would decrease Ca++ mobilization, and that crosslinking severely decreased Ca++ mobilization and eliminated clonal expansion (39). Apparently, the block to T cell activation that occurs on crosslinking CD4 stops cell differentiation at the G1/S phase, just prior to cell division (85). Further, the same proliferative defect was produced in cells from seronegative volunteers when pretreated with both gp120 and anti-gp120 antibody. None of these effects require an HIV infection, only the presence of gp120 and anti-gp120 antibody to crosslink surface CD4. Indeed, the studies of in vitro HIV infections use such artificially high levels of T₄ cell activation and viral infection that the physiologically relevant results would never be seen -- CD4 expression levels would be negligible and crosslinking could not occur.

More support for CD4 crosslinking as a cause of anergy comes from a study where restoration of the proliferative response of T₄ cells from HIV+ patients during in vitro culture paralleled the loss of gp120 IC from the T₄ cell surface (83). Further evidence comes from indirect in vivo studies, where a decrease in lymphocyte proliferation was directly associated with the immune complex load on T₄ cells (97). Again, the signal delivered by CD4 crosslinking is distinct from the effects of an HIV infection or of gp120's binding without crosslinking, and it offers the best explanation for the T₄ cells inability to proliferate.

Of all of the effects in AIDS, the most mysterious when viewed through the direct infection model is the change in cytokine hormones, the increase in IL-4 and -10 that follows the decrease in IL-2 (98). While there are no gp120 IC crosslinking experiments on this topic in the literature, there are several reports of CD4 crosslinking effects on cytokine production (47, 86, 92-94, 99). Most interestingly, while there is variation in the extent and type of response to CD4 crosslinking depending on the antibody epitope (84), nearly all crosslinking events cause an up-regulation of IL-4, most cause an increase in IL-10, and all severely reduce IL-2 levels. Until experiments are done to examine the effects of gp120 IC crosslinking on cytokines, we cannot know its role, but a causal relationship seems likely.

In modifying the cytokines present in an HIV+ patient, the immune complex crosslink automatically causes responding cell switching in the immune system. With the systemic overexpression of IL-4 and -10 (TH2 cytokines), the development of B cells is heavily favored over that of T₈ and natural killer cells. This leads to diminished clearance of cancers, intracellular parasites such as Listeria, Mycobacteria, Toxoplasmosis, Cryptosporidium, and viral pathogens (EBV, CMV, Herpes) (50, 93, 98, 100, 101). It is striking that this list contains most of the novel opportunistic parasites encountered during AIDS, with the exception of Pneumocystis carinii, whose kingdom affiliation and cellular location in the host are both under question (102). In controlled studies, inappropriate B cell activation and T₈ underutilization have been seen both with constituative overexpression of IL-4 and with administration of tolerance inducing anti-CD4 antibodies for transplantation (103, 104). Unfortunately, the bias toward B cell expansion at the cost of T₈ efficacy is self-reinforcing (92), and the immune system seems to lose its ability to balance the responding cell types appropriately after a few weeks (100). While this condition does not remedy itself, there is promise in restorative IL-2 therapy and the administration of anti-IL-4 antibodies for reinstating balance in the immune system (92, 99).

The prognosis for reversing anergy is not so positive. Unlike the reversible effects from cytokine imbalances or gp120 alone, the anergy produced by crosslinking CD4 is both persistent and neither preventable nor reversible by IL-2 (91, 95). This immediately points to a mechanism by which IL-2 dosing would be both dependant on the number of T₄ cells and would downregulate its own efficacy. With a larger pool of T₄ cell targets there is a greater probability of administering a dose of IL-2 to correct for gp120 binding effects before any specific cell is crosslinked, allowing for some small span of appropriate T₄ cell function before anergy sets in. On continual dosing with IL-2, the small pool of active T₄ cells provide helper functions for both the downregulated T₈ cells and the IL-4 stimulated B cells, leading to an increase in antigen-driven antibody production response and an increase in antibody titer. The antibodies that are specific for gp120 then increase gp120 immune complex crosslinking on T₄ cells and render them unresponsive to further IL-2 therapy. In cases where the number of infected cells and free viral particles is minimal, e.g., through the use of HIV suppressive drugs, the concentration of gp120 is also reduced and this can manifest as a partial recovery. Without gp120 the anti-gp120 antibodies cannot crosslink CD4 and are harmless. The reduction in free HIV results in less gp120 and thus in a reduction of CD4 crosslinking. Though heavily biased toward B cell responses, the patient would be expected to maintain a significant immune response. However, with a compromised T₈ system, any significant bout of illness (which increases HIV expression from previously latently infected cells) or development of viral drug resistance would lead to a rapid reversal to AIDS. Additionally, the small amount of crosslinking that is unavoidable would be expected to mimic long term low dose anti-CD4 therapy and produce "infectious tolerance". In infectious tolerance, mainly seen in transplant cases after immunosuppression, the tolerance to a particular antigen developed during immunosuppression is passed along to new generations of immune cells (and as a laboratory exercise, to other animals). This results in a persistent lack of response to previously encountered pathogens despite having an otherwise healthy and functioning immune system (92).

Apoptosis

Unlike anergy, apoptosis (programmed cell death, PCD) has received quite a bit of attention in the AIDS literature, as it is consistent with the current focus on T₄ cell deletion. While there are explanations for the PCD of HIV infected cells, readily acknowledged as part of the direct infection model, there is no direct infection model for "bystander" PCD. As the majority of cell death in an HIV+ patient is that of bystander cells this is not a satisfactory level of explanation. Again, immune complex crosslinking of CD4 provides a clear, consistent mechanism and a wealth of both precedent and experimental data.

From the standpoint of anti-CD4 antibody research, the induction of PCD is nothing new. In addition to directly inducing PCD by crosslinking CD4 before T cell receptor stimulation (88), crosslinking increases cell surface expression of the PCD regulatory molecule fas -- sensitizing the cell to signals for death (105). This fas enhancing aspect of crosslinking may contribute significantly to PCD, as antibodies are raised against a V3 epitope of gp120 that has homology to fas (106). These anti-fas antibodies can crosslink fas molecules and deliver the signal to initiate PCD. As an additional factor, the CD4 crosslinking effect on the balance of cytokine hormones appears to influence T₄ cell sensitivity to PCD. The cytokines suppressed by CD4 crosslinking (IL-2) ameliorate PCD and the cytokines enhanced by crosslinking (IL-4, IL-10) exacerbate PCD (15, 77).

Beyond parallels and supporting factors, there is direct evidence of T₄ cell deletion by CD4 crosslinking. In separate experiments it was confirmed that gp120 immune complexes but not gp120 itself, mediated PCD (107, 108). These experiments also found PCD induced by gp120 IC, like PCD induced by anti-CD4 antibodies, could not be prevented by coadministration of IL-2. Apparently T cells can be induced to undergo PCD simply by processing both a CD4 crosslinking signal and a T cell receptor signal that have been separately delivered. Given the large fraction of CD4 consistently bound by gp120 and immune complexes in an HIV+ patient, it seems likely that this contributes to the deletion of T₄ cells in vivo. These studies are consistent with several antibody and T₄ cell function studies: the patient's concentration of antibodies is negatively correlated to T₄ cell count (109); surface-bound antibody is negatively correlated to T₄ cell count (97, 110); and early phase high antibody titer is positively correlated with rate of progression to AIDS / ARC (111).

Further evidence of the ability of gp120 immune complexes to delete T₄ cells comes from an in vivo experiment using murine T₄ cells transgenic for human CD4 (112). By placing the full gene and regulatory elements for the functional expression of human CD4 in mice, a model was created where T₄ cells can bind gp120 IC and transduce their signals, yet not permit HIV infection. Experiments in this system clearly showed that dosing with gp120 immune complexes led to a specific and significant deletion of T cells bearing human CD4. The T cells were not deleted when either the gp120 or the anti-gp120 antibody was excluded. When the experiment was conducted in mice that had human CD4 expressed on their B cells, no cells were deleted. The B cells cannot transduce signals from CD4 and are therefore not affected by CD4 crosslinking. This specific in vivo deletion cannot be caused by HIV infection, as all of the cells are from mice and therefore uninfectable by HIV. It occurs with the presence of gp120 and anti-gp120 antibody, and nothing else is required. This experiment requires, at a minimum, an addition to the direct infection model, if not a thorough revision.

These PCD experiments thoroughly demonstrate the strengths of a gp120 IC mediated model of AIDS: cells that are infected by HIV supply gp120 to bind to the CD4 on neighboring cells. The bound gp120 is crosslinked through antibody produced by the host, and this initiates PCD of the "bystander" uninfected cells. The infected cells do not express CD4 on their surface and thus are not a target for antibody crosslinking. This explains their survival compared to the "bystander" cells, and they persist as a source of both virus and free gp120. Additionally, this model accounts for the previously paradoxical observation that AIDS only occurs after the onset of anti-HIV immune responses.

Other curious features of AIDS fall into place as well: resolution to the infection / death discrepancy as each virus carries enough gp120 to kill 100+ T₄ cells (and infected cells contribute still more), how qualitative defects appear while T₄ cell numbers are in normal ranges, the inability of IL-2 to reverse the effects of HIV infection once symptoms (crosslinking) have appeared, the cytokine shift is no longer mechanistically orphaned, and the differences in symptoms between pure T₄ cell culture and PBMC (which contain B cells) have a plausible explanation. The increase in numbers of HIV infected cells during the progression to AIDS has already been questioned by AIDS researchers: is it the cause or the result of the immune system collapse (113, 71)? It seems likely to be the result of an immune collapse caused by gp120 and gp120 immune complexes. Even the switch in viral tropism from macrophage to T₄ cell prior to the onset of AIDS, previously thought to support a direct infection model, also supports an immune complex model: The T cell tropic strains shed gp120 more readily, and this increases the free gp120 pool for CD4 binding. Nearly all of the features of AIDS can be explained through either direct gp120 effects or gp120 mediated crosslinking. Few are explainable through a direct infection model. It is more reasonable to believe that AIDS is caused by the simultaneous presence of gp120 and anti-gp120 antibodies; the role of HIV is merely to supply the gp120.

Clonal De-Selection and Immunotoxin Design

In addition to being comprehensive and well supported, the immune complex model offers a clear therapeutic approach. In an HIV infection, the normally beneficial antibody response causes more damage than benefit. If there were a way to stop the production of antibodies to gp120, while leaving the rest of the immune system intact, the progression of an HIV infection to full-blown AIDS could be slowed or even halted. There is even the possibility for a complete and durable reversal of AIDS symptoms in people who have already progressed into AIDS / ARC. While these people would still be HIV-positive, they should show few or no symptoms.

Fortunately, there is an established, clear, simple way to do this. Because B cells must internalize their target antigen in order to start making antibodies to it, this makes them easy targets for an "immunotoxin". If we take the material that the B cells internalize, in this case gp120, and attach a toxin molecule to it, then the B cells that pick up gp120 (6% of the total) will poison themselves. Cells that do not internalize gp120 will not be exposed to the toxin. Thus the B cells that react to gp120 will no longer exist to make antibodies to gp120. The concentration of anti-gp120 antibodies in the blood will fall, crosslinking of CD4 (and also anti-gp120 mediated fas crosslinking) will end, and the T₄ cells can be made to carry out their essential tasks by restoring normal levels of IL-2. The rest of the B cells will continue to make antibodies to other foreign materials, and the T₈ cells will continue to kill virus-infected cells, as they would normally. There is reason to believe this strong T₈ response will be able to control an HIV infection, as long as functional T₄ responses and normal IL-2 levels are maintained (1, 3, 4, 13, 22, 114).

This immunotoxin can be made by joining, preferably at the genetic level, any one of the many toxins that translocate from the lysosome to cytosol with gp120, which is the means of delivery to B cells. By creating and expressing the immunotoxin as a single fusion protein, any problems with premature / inappropriate disassociation of the toxin from the targeting sequence are automatically solved, and inefficient racemic chemical coupling is avoided. The candidate we intend to prepare and test is a receptorless Diphtheria toxin, truncated at amino acid (aa) 389. Diphtheria toxin (and the similar Pseudomonas exotoxin) are ideal candidates for these immunotoxins as they translocate easily and specifically from the lysosome to the cytosol, then enzymatically inactivate protein synthesis through ADP-ribosylation of EF-2 (115, 116). Because the Diphtheria toxin's normal receptor binding domain is located in the amino acids deleted from the toxin, aa's 525-535 (117), this truncation eliminates diphtheria toxin's ability to bind to its normal receptors and dramatically eliminates its nonspecific toxicity (118). It is potent enough to kill at a dose of one molecule per cell (119), yet without a way to get into cells, it is almost totally harmless. Amino acids 390 to 871 of the immunotoxin would be supplied by fusion of a modified gene for gp120. When the receptorless toxin is attached to the modified gp120 (which functions as a targeting sequence) the resulting toxin would target only those cells that internalize gp120 and no other cells in the body.

In order to prevent T₄ cells from internalizing the toxin, as they do gp120, the gp120 joined to Diphtheria toxin would be made incapable of binding to CD4. By mutating the gp120 sequence from 368D and 370E to 368K and 370R it should be possible to reduce the CD4 binding affinity by over five orders of magnitude (120). Expressing the immunotoxin in E. coli also guarantees production in a non-glycosylated form, which further decreases the CD4 affinity by 50 fold (121), though it does not change antibody or B also guarantees production in a non-glycosylated form, which further decreases the CD4 affinity by 50 fold (121), though it does not change antibody or B also guarantees production in a non-glycosylated form, which further decreases the CD4 affinity by 50 fold (121), though it does not change antibody or B cell recognition (122). In this manner it should be possible to restrict the toxin internalization to the B cells that are reactive to gp120, the ones specifically targeted for deletion.

The immunotoxin approach is not new, only new to this target. Several immunotoxins have gone from laboratory studies (115, 123, 124) to full scale clinical trials for cancer treatments, many using the same Diphtheria toxin fragment (125). A component of these studies is a toxicity profile, and it was found that an IL-2 based immunotoxin of similar design dropped from a LD50 of 150 ng/kg for wild type Diphtheria toxin to 3 mg/kg for the immunotoxin -- a >20,000 fold reduction in whole animal toxicity (115). Considering that IL-2 receptors are more numerous and widely distributed through the body than our target cells, this suggests a low toxicity and lower required dose for our proposed toxin. In terms of delivery efficacy to the sensitive cells this IL-2 toxin was found to have an IC50 of 1x10-10M, a level comparable to other well designed Diphtheria-based immunotoxins (118, 126). While many immunotoxins have had some measure of difficulty in penetrating solid tumors and effectively dosing target cells, the vascular nature of lymph nodes and the peripheral blood circulation of B cells guarantee accessibility of our targets and ease of treatment.

Additionally this immunotoxin could easily be updated to remain effective on new mutant forms of gp120, simply by isolating and grafting the new gp120 sequence to the toxin cassette. Furthermore, since the immunotoxin is targeted at human cells rather than the mutation prone virus, HIV cannot develop its characteristic drug resistance. This resistance is already being seen with the nucleotide analogs and protease inhibitors. With the small-molecule drugs designed to inhibit viral processes any significant mutation in the binding site of the drug's target enzyme system requires redesigning the drug to fit the new geometry. Redesigning the drug molecule often requires redesigning the synthesis, altering the factory conditions for its production, and another round of FDA testing as the new drug may be metabolized differently or show new side effects. The current generation of protease inhibitors seems to enhance the risk of diabetes, what other regulatory processes might be impacted by improper protein degradation? In contrast, the same PCR methods used to assess viral burden can generate copies of the gp120 genes in an individual, use of these genes in immunotoxin construction is straightforward and simple. The updated immunotoxins would be metabolized like the original toxin, and all other proteins, and the inability of the toxin to enter non-selected cells dramatically reduces the risk of crossreaction with normal human enzymes and cells. There is little reason to expect different properties or side-effects from an updated immunotoxin, while there is a large risk with both protease inhibitors and nucleotide analogs. Also, the infectivity of people treated by this drug should be low, characteristic of people in the asymptomatic and relatively immunocompetent stage of an HIV infection.

The immunotoxin we are proposing uses the immunologic interaction of gp120 to target an extremely potent toxin to a small and specific population of B cells. This removes the CD4 crosslinking gp120 antibody response and leaves the rest of the body untouched. With this treatment, we believe that the collapse of the immune system characteristic of AIDS can be held off indefinitely in people who are HIV-positive.

References

Please see the References page for a complete listing of the references used in this article.

Acknowledgements

I gratefully acknowledge Elizabeth Andrews, Frank Andrews, G. George Capps, Shirley Fink, Beth Trumbo, the Santa Cruz AIDS Research Foundation, and all the people who have supported this project.

Support Our Research     Contact Us     Technical Information