Autologous tumor vaccines processed to express α-gal epitopes: a practical approach to immunotherapy in cancer

Abstract

This review describes a method by which the human natural anti-Gal antibody can be exploited as an endogenous adjuvant for targeting autologous tumor vaccines to antigen-presenting cells (APCs). Tumor cells remaining in the patient after completion of surgery, radiation, and chemotherapy are the cause of tumor relapse. These residual tumor cells can not be detected by imaging, but their destruction may be feasible by active immunotherapy. Since specific tumor-associated antigens (TAAs) have not been identified for the majority of cancers, irradiated autologous tumor vaccines have been considered as an immunotherapy treatment that may elicit an immune response against the residual tumor cells expressing TAAs. However, tumor cells evolve in cancer patients in a stealthy way, i.e., they are not detected by APCs, even in the form of vaccine. Effective targeting of tumor vaccines for uptake by APCs is a prerequisite for eliciting an effective immune response which requires transport of the vaccine by APCs from the vaccination site to the draining lymph nodes. In the lymph nodes, the APCs transporting the vaccine process and present peptides, including the autologous TAA peptides for activation of the tumor-specific T cells. The required targeting of vaccines to APCs is feasible in humans by the use of anti-Gal. This antibody interacts specifically with the α-gal epitope (Galα1-3Galβ1-4GlcNAc-R) and is the only known natural IgG antibody to be present in large amounts in all humans who are not severely immunocompromised. The α-gal epitope can be synthesized on any type of human tumor cell by the use of recombinant α1,3galactosyltransferase (α1,3GT). Solid tumors obtained from surgery are homogenized and their membranes subjected to α-gal epitope synthesis. Similarly, α-gal epitopes can be synthesized on intact tumor cells from hematological malignancies. Administration of irradiated autologous tumor vaccines processed to express α-gal epitopes results in in situ opsonization of the vaccinating cells or cell membranes due to anti-Gal binding to these epitopes. The bound antibody serves to target the autologous tumor vaccine to APCs because the Fc portion of the antibody interacts with Fcγ receptors on APCs. Since patients receive their own TAAs, the vaccine is customized for autologous TAAs in the individual patient. The repeated vaccination with such autologous tumor vaccines provides the immune system of each patient with an additional opportunity to be effectively activated by the autologous TAAs. In some of the immunized patients this activation may be potent enough to induce an immune-mediated eradication of the residual tumor cells expressing these TAAs.

Autologous tumor vaccines as an immunotherapy treatment modality

The objective of this review is to present a method for eliciting a tumor-specific immune response in cancer patients by using the autologous tumor as a vaccine that has been processed so that it is effectively targeted in situ to antigen-presenting cells (APCs). This targeting is mediated by the natural anti-Gal antibody (Ab) that constitutes ~1% of circulating IgG in humans. Cancer patients relapse and develop lethal disease because of tumor cells that remain as micrometastases, or as minimal residual disease after completion of surgery, chemotherapy, and radiation therapy. It may be possible to eradicate these tumor cells by active immunotherapy that elicits an effective immune response against tumor-associated antigens (TAAs) expressed on the tumor cells. The major unknown element in cancer immunotherapy is the nature of TAAs to be used as vaccine. Research in melanoma has demonstrated the induction of an immune response against a variety of characterized TAAs, as reviewed elsewhere [1–5]. In addition, over-expressed normal antigens (Ags) [2, 6–9] and viral Ags [2, 10–12] have been found on various tumors and may serve as targets for immunotherapy. However, in most types of cancer, in particular in those with a high recurrence rate (e.g., pancreatic and renal carcinomas), no specific immunotherapy treatment is available to patients, since the identity of the TAAs is unknown. In addition, immunotherapy studies with characterized TAAs have indicated that immunization with a single type of TAA molecule may not suffice due to immunologic selective pressure for the appearance and expansion of tumor cells with low or no expression of the specific TAA [13–15]. Therefore, effective tumor vaccines are thought to require the inclusion of several TAAs—i.e., polyvalent tumor vaccines [15].

In the absence of current sufficient information on TAAs, immunotherapy is not included as a treatment modality in many types of cancer. An alternative immunotherapy approach which does not require TAA identification uses the autologous tumor as a source for vaccinating TAAs. One such method involves the in vitro pulsing of dendritic cells (DCs) with peptides from the autologous tumor, followed by immunization of the patient with the pulsed DCs [16–21]. Another method includes the pulsing of autologous DCs with RNA extracted from the tumor [22]. The efficacy of in vitro–pulsed DCs as vaccine in humans is presently under evaluation. However, the in vitro culturing of a large number of the patient’s DCs for the repeated vaccination, the isolation of the eluted peptides, and the in vitro pulsing of DCs with the peptides, all require specialized resources, limiting phase II/III studies to only a few centers. Moreover, because of technical limitations in generating human DCs in vitro in large enough amounts, the number of pulsed DCs administered into patients per body weight is 100–200-fold lower than that used in mice for successful antitumor immunization (i.e., ~1×10⁶ pulsed DCs per mouse vs ~1×10⁷–2×10⁷ pulsed DCs per patient, while humans are 2,000–3,000 times larger than mice) [16–19, 23, 24]. An alternative, more feasible approach for active immunotherapy in a large number of cancer patients is direct immunization with irradiated autologous tumors that have been processed to be immunogenic. One method studied in melanoma patients has been the linking of the hapten dinitrophenol (DNP) to autologous tumor cells. This method was reported to increase immunogenicity of autologous melanoma cells used as vaccine [25, 26].

Autologous tumor vaccines require effective targeting to APCs

The development of tumors may be regarded as a failure of the immune system to identify the tumor cells as cells that must be destroyed. This failure in immune surveillance against the tumor has been attributed to a variety of factors [13, 14, 27], including (1) down-regulation of TAA expression on tumor cells, (2) loss of HLA class I expression on the tumor cells, (3) defective apoptosis pathways, (4) immunosuppressive cytokines, and (5) poor uptake and processing of tumor cells by APCs because of failure to detect the tumor cells as cells that have to be internalized and their TAAs processed. Whereas factors 1–4 are difficult to control and modify in a clinical setting, this review presents a method by which any type of autologous tumor vaccine in humans can be modified so it will be effectively internalized and processed by APCs.

When irradiated autologous tumor cells or cell membranes are injected intradermally as a vaccine, they must be internalized by APCs which transport the vaccine to the draining lymph nodes [24, 28–30]. Within the lymph nodes, the APCs process the internalized vaccine and present peptides, including the TAA peptides, in association with MHC class I and class II molecules for activation of the corresponding CD8⁺ and CD4⁺ T cell, respectively. Once the TAA-specific T cells are activated, they can leave the lymph nodes, circulate in the body, and seek malignant cells expressing the TAA in order to destroy them. Abs against the TAA may also be produced in this process, as a result of helper T-cell activation. Such Abs contribute to the destruction of metastatic cells by inducing Ab-dependent cell-mediated cytolysis (ADCC) of the tumor cells [23, 31, 32]. In some patients, the combined cellular and humoral immune response may be strong enough to eradicate the tumor cells in micrometastases and those of the minimal residual disease in hematological malignancies. However, since the effective uptake of tumor cells by APCs is a prerequisite for the immune destruction of the developing tumor, tumor cells in cancer patients evolve under a constant selective pressure to be “invisible” to APCs [13]. In addition, tumor cells may not express “danger signals” which are required for identification of a particulate material as an antigenic entity that is to be processed by APCs [33]. Therefore, autologous tumor vaccines are also likely to be unrecognized by APCs, as cells or membranes that should be internalized, transported to the lymph nodes, and processed. Even if APCs are recruited to the vaccination site by cytokines such as GM-CSF [34–37], the autologous tumor vaccine may still not be recognized by the APCs as particulate material that has to be internalized. Thus, the uptake of such vaccines may be poor, since it is mediated only by random accidental endocytosis [23]. In order for autologous tumor vaccines to succeed, they must be effectively targeted in situ to APCs. This can be achieved by exploiting the expression of Fcγ receptors (FcγRs) on APCs.

Targeting vaccines to Fcγ receptors of APCs enhances immunogenicity

In vivo targeting of various vaccines to APCs can be achieved by complexing with IgG molecules (i.e., opsonization). This is because APCs, including macrophages, Langerhans cells of the skin, and DCs, all express FcγRs that can effectively induce uptake of IgG-complexed Ags, following interaction with the Fc portion of the opsonizing IgG molecules [32, 38–40]. Moreover, binding of Ag-Ab immune complexes to FcγRs of DCs induces effective maturation of the DCs, resulting in effective cross-presentation (cross-priming) of the antigenic peptides by MHC class I molecules for the activation of CD8⁺ cytotoxic T cells, and presentation of peptides on MHC class II molecules for the activation of CD4⁺ helper T cells [23, 41–44]. The principle of increasing immunogenicity of vaccines by complexing them with their Abs has been demonstrated with a variety of Ags, including: tetanus toxoid [45–47], hepatitis B Ag [48, 49], and Eastern equine encephalomyelitis virus [50]. Immune complexes between simian immunodeficiency virus (SIV) and anti-SIV Abs were also found to be targeted to APCs, resulting in enhanced cross-presentation of SIV peptides [51]. Moreover, cytotoxic T cells in the SIV-infected monkeys were effectively activated by MHC class I–presented peptides following immunization with SIV–anti-SIV immune complexes and this effect was abrogated when the Fc portion of the anti-SIV Abs was enzymatically removed [51]. Similarly, natural Abs in mice were found to function as an endogenous adjuvant forming immune complexes with Leishmania vaccine and to induce a strong CD8⁺ T-cell response against the intracellular form of the parasite [52]. A protective effect of immune complexes targeted to DCs was also demonstrated in an experimental model of cancer immunotherapy [23]. Mice, challenged with tumor cells expressing ovalbumin (OVA) as a surrogate TAA, displayed a much more effective immune protection if they received DCs that were pulsed with OVA–anti-OVA immune complexes, than if the DCs were pulsed only with OVA. This increased immune protection was associated with elevated activation of OVA peptide–specific CD4⁺ and CD8⁺ T cells, as well as with increased production of anti-OVA Abs. Furthermore, this immunoprotective effect of pulsing with the OVA–anti-OVA immune complexes could not be observed if the DCs lacked FcγR (i.e., DCs from FcγR knockout mice) [23]. All these studies imply that complexing of autologous tumor vaccines with an IgG Ab is likely to mediate the required effective in vivo targeting of the vaccine to APCs. Since humans (except for severely immunocompromised individuals) produce large amounts of the natural anti-Gal Ab [53], this Ab can be exploited as an endogenous adjuvant enhancing immunogenicity of intradermally administered autologous tumor vaccines, by targeting such vaccines to APCs [54].

The natural anti-Gal Ab, the α-gal epitope, and targeting of tumor vaccines to APCs

Anti-Gal is the only natural IgG Ab that is known to be present in all humans in large amounts, constituting ~1% of serum IgG (20–100 μg/ml serum) [53]. Anti-Gal interacts specifically with α-gal epitopes (Galα1-3Galβ1-4GlcNAc-R) on cell surface glycolipids and glycoproteins [55, 56], and it is produced throughout life as a result of antigenic stimulation by bacteria of the gastrointestinal flora [57]. As many as 1% of B lymphocytes in humans are capable of producing this Ab [58]. Cancer patients produce anti-Gal as well as healthy individuals, unless they are severely immunocompromised [53, 59]. In our experience, humans with physiologic serum IgG concentration (7–12 mg/ml) display normal levels of anti-Gal activity [53]. Anti-Gal activity can be readily measured by ELISA with glycoproteins expressing α-gal epitopes as solid phase Ag, such as synthetic α-gal epitopes linked to bovine serum albumin, or with mouse laminin [58, 59]. Individuals with suppressed IgG production because of the lack, or insufficient activity, of normal B cells, or patients with advanced myeloma, where the proportion of physiologic IgG is greatly reduced, display decreased titers of anti-Gal [53]. The α-gal epitope which is the specific ligand for anti-Gal, is absent in humans, but is abundantly synthesized on glycolipids and glycoproteins by the glycosylation enzyme α-1,3-galactosyltransferase (α1,3GT) within the Golgi apparatus of cells of nonprimate mammals, prosimians, and in New World monkeys [60, 61]. In contrast, humans, apes, and Old World monkeys lack α-gal epitopes, but produce the anti-Gal Ab [60, 61]. The absence of α-gal epitopes in humans, apes, and Old World monkeys is the result of inactivation of the α1,3GT gene in ancestral Old World primates, approximately 20–30 million years ago [62–66].

Anti-Gal is an Ab that can be very active in vivo, as can be inferred from studies in xenotransplantation. In vivo binding of anti-Gal to α-gal epitopes on transplanted pig heart or kidney is the main cause for the rapid rejection of such grafts in humans and in Old World monkeys [67–70]. Similarly, it is possible to hypothesize that autologous tumor cells or membranes processed to express α-gal epitopes will bind the natural anti-Gal IgG Ab in situ at the vaccination site [54, 71]. Such interaction is likely to target the vaccine to APCs (Fig. 1). The binding of the Fc portion of the complexed anti-Gal to FcγRs on APCs will induce effective internalization of the opsonized vaccinating tumor cells into APCs. Thus, the uncharacterized TAAs of the autologous tumor will also be internalized. After transport of autologous tumor vaccines to the draining lymph nodes, the APCs will process and present the TAA peptides for activation of tumor-specific cytotoxic and helper T cells.

Fig. 1.

Anti-Gal–mediated targeting of autologous tumor vaccines expressing α-gal epitopes to APCs. Anti-Gal of the vaccinated patient binds in situ to α-gal epitopes on vaccinating tumor cells or cell membranes. The subsequent interaction between the Fc portion of the bound anti-Gal and Fcγ receptors (FcγRs) on APCs induces phagocytosis of the vaccine into the APCs and thus the effective internalization of the tumor-associated antigens (TAAs) on the tumor membranes. The internalized TAAs are processed and the various immunogenic TAA peptides (circle, square, and triangle) are presented by the APCs in association with MHC class I or II molecules

The ability of anti-Gal to increase uptake, processing, and presentation by APCs could be demonstrated by the activation of a helper T-cell clone specific for a peptide of the influenza virus hemagglutinin [72]. Mouse APCs that were incubated with influenza virus expressing α-gal epitopes and complexed with anti-Gal were tenfold more effective in activating helper T cells specific for the processed viral peptides than APCs incubated with the virus lacking α-gal epitopes and thus not binding anti-Gal [72]. Human anti-Gal IgG readily interacts also with FcγRs on human APCs. This can be inferred from observations in xenotransplantation where anti-Gal IgG bound to α-gal epitopes on pig cells, further interacts with FcγRs on human macrophages and NK cells and stimulates them to exert their cytotoxic potential by the ADCC mechanism [67, 73]. Based on these observations, it was of interest to develop a method for processing human tumor cells to express α-gal epitopes, and to determine whether such cells can be opsonized by anti-Gal and subsequently internalized by human APCs.

Processing human tumor cells to express α-gal epitopes

To express α-gal epitopes on autologous tumor cells or cell membranes the α1,3GT gene was cloned from a marmoset cell line (marmoset is a New World monkey that is evolutionarily closer than other nonprimate mammalian species to humans) [74], using the information on the sequence of this gene in mice [75]. The cloned gene was truncated to remove the transmembrane and cytoplasmic domains, to allow secretion of the recombinant (r) α1,3GT, and to prevent enzyme aggregation. This recombinant enzyme could be readily expressed in E. coli [76], or in mosquito cells infected with baculovirus containing the α1,3GT gene [71]. However, since these two expression systems involve isolation of the recombinant enzyme from cell lysates, there was a possibility that the enzyme would not be of sufficient purity for processing autologous tumor vaccines that are to be administered into cancer patients. Therefore, the rα1,3GT production was adapted to the yeast Pichia pastoris expression system [77] where the recombinant enzyme is secreted into the culture medium. The enzyme with a fused (His)₆ tag can be readily purified by a nickel-Sepharose affinity column. The purified rα1,3GT was found to be very effective in synthesis of α-gal epitopes on asparagine (N)-linked carbohydrate chains of glycoproteins with a specific activity of 8×10¹¹ epitopes/min/μg (i.e., ~1.2 nM/min/mg) [77].

A large proportion of carbohydrate chains of human glycoproteins and glycolipids have the core structure of N-acetyllactosamine (Galβ1-4GlcNAc-R) that is capped with sialic acid (SA) i.e., SA-Galβ1-4GlcNAc-R (left structure in Fig. 2). The synthesis of α-gal epitopes on glycoproteins and glycolipids is achieved in a two-step reaction [54, 59, 71, 76, 78]: (1) SA is removed by neuraminidase (1 mU/ml, commercially available) in order to expose the penultimate N-acetyllactosamine residues, and (2) rα1,3GT (20–50 μg/ml) synthesizes α-gal epitopes on these N-acetyllactosamine residues by linking galactose that is contributed by the sugar donor uridine diphosphate galactose (UDP-Gal, 1 mM). Step 2 of this enzymatic reaction is similar to the natural synthesis of α-gal epitopes within the Golgi apparatus in cells of nonprimate mammals, prosimians, and New World monkeys. The reaction is performed for 2 h at 37°C in a pH 6.2 buffer which contains both enzymes as well as Mn⁺⁺ ions. These ions are required for the catalytic activity of rα1,3GT. By using this reaction one can synthesize α-gal epitopes on >1×10⁹ human tumor cells in a reaction volume of 10 ml [78]. Quantification of the α-gal epitopes synthesized on tumor cells demonstrated the expression of 0.5×10⁶–10×10⁶ α-gal epitopes per cell, whereas the original unprocessed tumor cells completely lacked this epitope (unpublished observations). The variations in the number of α-gal epitopes per cell are likely to be associated with the type of carbohydrate chains on cell surface glycolipids and glycoproteins and with the surface area of the different tumor cells. The larger the cell, the more carbohydrate epitopes are available for conversion into α-gal epitopes. Since all human normal and malignant cells express glycolipids and glycoproteins with the structures of SA-Galβ1-4GlcNAc-R, or Galβ1-4GlcNAc-R, it is probable that the cell membranes of any type of human tumor can be processed to express α-gal epitopes. This reaction has no effect on TAA protein molecules or on TAA peptides since it modifies only existing carbohydrate chains on the cell membrane.

Fig. 2.

Processing of tumor cell membranes to express α-gal epitopes on asparagine-linked (N-linked) carbohydrate chains of glycoproteins. Synthesis of α-gal epitopes is achieved in a suspension containing neuraminidase and recombinant α-1,3-galactosyltransferase (rα1,3GT) in a two-step reaction: (1) removal of sialic acid (SA) from carbohydrate chains by neuraminidase to expose the penultimate N-acteyllactosamine residues (Galβ1-4GlcNAc-R), and (2) linking of galactose from the sugar donor uridine-diphosphate-galactose (UDP-Gal) to the exposed N-acetyllactosamine residues by rα1,3GT to form α-gal epitopes. The de novo synthesized α-gal epitopes readily bind the natural anti-Gal Ab

Whereas it is relatively simple to obtain a large number of tumor cells for vaccine preparation in patients with hematological malignancies, it is very difficult to obtain sufficient number of vaccinating cells from most solid tumors. A practical approach for preparation of autologous tumor vaccines in patients with solid tumors is to homogenize the tumor obtained from the surgery, wash the tumor membranes, and subject them to the enzymatic reaction. Effective synthesis of α-gal epitopes could be demonstrated on membrane homogenates from human mammary, colon, and ovarian carcinomas [54, 59]. The tumor membranes can be incubated at a concentration as high as 100 mg/ml with neuraminidase and rα1,3GT. Analysis of these tumor membranes following the enzymatic reaction indicated that they express ~2×10¹¹ α-gal epitope per mg of membranes [59]. Western blot analysis of the processed tumor membranes stained with human anti-Gal, indicated that α-gal epitopes are expressed on many different cell surface glycoproteins, ranging in size from 15 to 200 kDa [59]. This implies that the rα1,3GT is not selective in its activity and it can synthesize α-gal epitopes on any glycoprotein that expresses N-acetyllactosamine residues as part of its carbohydrate chains.

After completion of the incubation with the two enzymes, both intact tumor cells from hematological malignancies and tumor membrane homogenates from solid tumors are washed to remove the enzymes. The processed tumor is stored frozen in aliquots until the patient reaches remission and is ready to be vaccinated with the processed autologous tumor vaccine. Expression of α-gal epitopes on the vaccinating material is confirmed by ELISA with the monoclonal anti-Gal Ab, M86 [79]. For this purpose, the tumor cells or membranes are dried in ELISA wells, resulting in their strong adhesion to the wells as solid-phase Ag. After blocking of the wells, binding of monoclonal anti-Gal is measured with these tumor membranes. Unprocessed tumor membranes do not bind the Ab, whereas membranes processed to express α-gal epitopes readily bind it [59].

It should be stressed that whereas the expression of α-gal epitopes on tumor homogenates requires the use of rα1,3GT, in intact tumor cells, such as in hematological malignancies, expression of this epitope may be achieved also by an alternative method of transduction with a replication defective adenovirus containing the α1,3GT gene. Effective expression of α-gal epitopes was demonstrated on HeLa cells transduced with this vector [80]. In addition, it is possible that improved electroporation methods may enable future transfection of a large proportion of human tumor cells with a plasmid containing the α1,3GT gene.

Anti-Gal–mediated phagocytosis of tumor cells by APCs

Targeting of tumor cells expressing α-gal epitopes to APCs could be demonstrated in vitro with both lymphoma and leukemia cells [71, 78]. For example, lymphoma cells obtained from lymph nodes of B-cell lymphoma patients were processed to express α-gal epitopes (as in Fig. 2), and studied for anti-Gal–mediated phagocytosis by human macrophages and by DCs cultured from peripheral blood mononuclear cells [78]. The tumor cells were incubated with a subphysiologic concentration of natural anti-Gal (10 μg/ml) in the presence of these APCs. After a 2-h incubation at 37°C, the APCs were washed and stained. As shown in Fig. 3, the anti-Gal opsonized lymphoma cells were effectively internalized by macrophages and DCs, whereas in the absence of α-gal epitopes (i.e., in untreated lymphoma cells), no phagocytosis was observed [78]. Accordingly, in the absence of anti-Gal, cells expressing α-gal epitopes were not internalized by APCs. Similar anti-Gal–mediated effective uptake was observed with leukemia cells [71] or with normal human red cells processed to express α-gal epitopes [71]. These findings imply that tumor membrane homogenates obtained from solid tumors are likely to undergo a similar anti-Gal–mediated targeting to APCs after they are processed to express α-gal epitopes. These observations further imply that this method for targeting tumor vaccines to APCs may also complement the method of recruiting APCs to the vaccination site by administration of GM-CSF with the vaccine [34, 35, 81, 82]. This is because the anti-Gal opsonized vaccinating tumor cells or membranes will be effectively identified as particles that ought to be internalized by the many APCs that migrate to the vaccination site because of the recruiting effect of GM-CSF.

Fig. 3A–D.

Anti-Gal–mediated targeting of freshly obtained B-lymphoma tumor cells by APCs. The lymphoma cells were processed to express α-gal epitopes and incubated with anti-Gal (10 μg/ml) and with macrophages (B) or dendritic cells (D). The elongated nucleus in (B) is that of the macrophage, and in (D) is that of the dendritic cell, whereas the round nuclei in (B) and (D) are of the lymphoma cells. The representative macrophage internalized five lymphoma cells and the dendritic cell internalized one lymphoma cell. Control lymphoma cells that were not processed to express α-gal epitopes and which were incubated with anti-Gal, were not internalized by macrophages (A), or dendritic cells (C) (×1,000). Reproduced from [78] with the publisher’s permission

Studies on autologous tumor vaccine in an experimental animal model

The in vivo efficacy of tumor vaccines expressing α-gal epitopes can not be studied in a regular experimental animal model. Unlike humans, mice and rats produce α-gal epitopes, lack anti-Gal, and thus can not simulate the relevant human immune parameters. Monkeys, while producing anti-Gal, can not be used as an experimental model since there are no autologous tumor cell lines in primates. The efficacy of these tumor vaccines could be demonstrated, however, in the experimental model of knockout (KO) mice for α1,3GT that are challenged with the highly tumorigenic mouse melanoma cell line BL6 [83]. These KO mice, which originate from C57BL/6 mice, lack α-gal epitopes because of targeted disruption of the α1,3GT gene [84]. Since these mice lack the α-gal epitope, they are not immunotolerant to it. Therefore, the KO mice are capable of producing anti-Gal in titers comparable to those in humans, when immunized with xenogeneic membranes expressing α-gal epitopes, such as rabbit red cell membranes [83, 85]. The BL6 melanoma cells originate from C57BL/6 mice, and are derived from B16 melanoma. BL6 cells are uniquely suited for this study because these tumor cells lack α-gal epitopes due to inactivation of the α1,3GT gene [86]. In contrast, most mouse tumor cell lines express α-gal epitopes, as well as normal wild-type mouse cells. Administration of live tumor cells that naturally express α-gal epitopes into KO mice producing anti-Gal, results in destruction of these cells by the Ab, similar to the anti-Gal–mediated rejection of xenografts expressing α-gal epitopes [87]. The tumorigenicity of the BL6 melanoma cell line is so high that no immune protection against challenge with live tumor cells could be achieved by vaccination of C57BL/6 mice with irradiated BL6 cells secreting cytokines such as GM-CSF and IL2 [88].

Knockout mice producing anti-Gal in titers similar to those in humans, were immunized with 2×10⁶ irradiated BL6 melanoma cells engineered to express α-gal epitopes [83]. Subsequently, the mice were challenged with live parental BL6 cells, which lack α-gal epitopes. Control mice were immunized with irradiated BL6 cells lacking α-gal epitopes and also challenged with live BL6 cells. When challenged with 2×10⁵ BL6 cells, the mice immunized with BL6 cells expressing α-gal epitopes displayed a threefold higher resistance to the BL6 challenge than the control group (63% vs 20%, respectively) [83]. When the mice received a larger challenge of 5×10⁵ BL6 cells, all mice in the control group developed tumors within 21–25 days, whereas a third of the mice immunized with BL6 cells expressing α-gal epitopes did not develop tumors following the challenge [83]. In the remaining two thirds of mice of this group, histological examination indicated that developing tumors were surrounded by inflammatory mononuclear cells, most of which were T lymphocytes and macrophages [83]. In contrast, tumors from the control group vaccinated with unmodified tumor cells, lacked inflammatory cells, implying that no immune response developed following the tumor challenge. These observations implied that the in situ binding of anti-Gal IgG to vaccinating tumor cells that expressed α-gal epitopes indeed increased the immunogenicity of these tumor cells. Thus, irradiated tumor cells expressing α-gal epitopes elicited a protective antitumor immune response even in mice that failed to destroy all tumor cells in the challenge due to the large tumor burden.

Tumor vaccines expressing α-gal epitopes do not elicit autoimmune response to normal Ags

One possible concern that may be raised in regard to autologous tumor vaccines is the presence of normal Ags, either on tumor cells or on normal cells within the vaccine. It could be argued that this vaccination procedure may cause breakdown in tolerance and induce an autoimmune response to these normal Ags. This issue was studied in KO mice producing anti-Gal that were repeatedly vaccinated (five times) with 100 mg syngeneic kidney or liver membranes processed to express α-gal epitopes. Two months after the completion of the immunization protocol, the kidneys and liver of the immunized mice were subjected to histologic analysis. No induction of autoimmune reactions within these organs could be detected, as indicated by the normal structure of the tissues and the complete lack of inflammatory cell infiltrates within the liver or kidneys of the immunized animals (unpublished observations). Moreover, the sera of the immunized mice contained no Abs to KO mouse liver or kidney membranes, as indicated by ELISA with such membranes as solid-phase Ags. Similar studies were also performed in KO mice producing anti-Gal that were repeatedly immunized five times with 100 mg of kidney or liver membranes obtained from the wild-type C57BL/6 mice, or with 20×10⁶ C57BL/6 lymphocytes, all expressing α-gal epitopes [79]. The mice were studied for autoimmune responses, 2 months after the last immunization. These immunizations also did not cause breakdown of tolerance, as indicated by the lack of auto-Abs, absence of any inflammatory processes in the organs of the immunized mice, and no lymphopenia (unpublished observations). These findings suggest that autologous tumor vaccines expressing α-gal epitopes are unlikely to induce an autoimmune response against normal tissue Ags in the vaccine. Nevertheless, in view of the vitiligo observed in some melanoma patients undergoing immunotherapy treatment [25, 26, 37], patients vaccinated with autologous tumor vaccines should be closely monitored for autoimmune responses.

Considerations regarding immunization of cancer patients

Based on the above observations, the FDA has approved phase I studies on autologous tumor vaccines in patients with hematological malignancies, ovarian carcinoma, and pancreatic adenocarcinoma (IND 11183 and IND 9685). These studies, which are aimed to determine toxicity of the vaccine in a dose-escalation manner, are to be carried out in patients with recurring disease. Participating patients with solid tumors are planned to receive six vaccines, each containing 20-, 40-, or 80 mg autologous tumor membranes processed to express α-gal epitopes. All autologous tumor vaccines will be irradiated with 50 Gy prior to injection in order to ensure the complete elimination of live cells. The vaccines will be given in 2-week intervals. Patients with recurring hematological malignancies are planned to undergo a similar vaccination protocol, receiving 10×10⁶, 50×10⁶, or 200×10⁶ tumor cells processed to express α-gal epitopes. There is no apparent reason to expect that these autologous tumor vaccines will be toxic, since the vaccinating tumor differs from the original tumor only by lacking sialic acid and expressing α-gal epitopes. Previous studies on vaccination of leukemia patients with irradiated autologous tumor cells treated with neuraminidase, reported no adverse effects of the vaccinating cells, despite the absence of sialic acid [89]. Similarly, abundant expression of α-gal epitopes on implanted pig valves or on fetal pig islet cells was found to exert no adverse effects on patients receiving these pig tissues [90]. In addition, α-gal epitopes are present in large amounts in cow and pork meat.

If phase I studies prove to be successful (i.e., no toxicity observed), phase II studies will evaluate the effect of autologous tumor vaccines in patients in remission. Patients with solid tumors who undergo surgery will subsequently receive the chemotherapy and/or radiation therapy that are currently used as treatment protocol. A few weeks after completion of therapy, the recuperation of the immune system will be confirmed by demonstrating a normal anti-Gal activity in the serum of the patient. As in phase I, the immunization protocol with the autologous tumor vaccine is planned to include six vaccinations (2-week interval between injections) with the autologous tumor membranes processed to express α-gal epitopes. Each vaccine will contain a maximum of 80 mg of tumor membranes, so that at the end of the vaccination protocol the patients will have received ~0.5 g of tumor membranes as vaccine. This amount corresponds to ~1.5 g of the original tumor tissue prior to homogenization. It is probable that if this amount of vaccine does not elicit an antitumor immune response, then even higher amounts of vaccinating membranes will have no effect on the immune system.

In patients with hematological malignancies, the amounts of isolated tumor cells may vary significantly from 30×10⁶ in myeloma patients to 2×10⁹ in CLL patients undergoing leukapheresis. It is suggested that in phase II, patients in remission will receive six vaccinations, each with processed autologous tumor cells, according to the amount of cells obtained from the individual patient. Such injections may contain 3×10⁶–2×10⁸ cells per vaccine. Subsequent to vaccination, the immune response to the tumor can be monitored by several analyses: (1) Humoral immune response is evaluated by measuring antitumor Abs in sera, assayed by ELISA with tumor membranes as solid-phase Ag. (2) Cellular immune response is evaluated by ELISpot assays with autologous tumor membranes for determining T_H1- and T_H2-cell activation. (3) In vivo immune response is evaluated by skin reaction of delayed-type hypersensitivity to injected tumor membranes. Such analyses will be correlated with monitoring of tumor development by standard clinical monitoring methods specific to each type of tumor. To determine whether such vaccination elicits an autoimmune response, auto-Ab production can be measured by ELISA with the corresponding cell membranes from normal tissue homogenates, as solid-phase Ags. Production of auto-Abs can be assessed by comparing Ab-binding at various serum dilutions in sera, prevaccination and postvaccination. By performing such phase II and later phase III studies, it will be possible to determine in what proportion of vaccinated patients and in which types of cancer, autologous tumor vaccines expressing α-gal epitopes induce effective eradication of the tumor cells that remain after surgery, radiation therapy, and chemotherapy.

Concluding remarks

The inability of currently used treatment protocols to effectively eliminate all tumor cells in many cancer patients, raises the need for intervention with active immunotherapy treatment modalities in order to achieve complete remission. Since for most types of cancer the identity of the TAAs is not known, the autologous tumor may be used as a source of vaccinating material. To elicit an effective antitumor immune response, tumor vaccines have to be targeted to APCs at the vaccination site, and transported by the APCs to the draining lymph nodes where the TAA peptides will be processed and presented for activation of tumor-specific T cells. Targeting of vaccines to APCs can be achieved by complexing them with IgG Abs which subsequently interact with FcγRs of APCs. The only known IgG Ab which is present in all humans and thus can be exploited for such targeting is the natural anti-Gal Ab. Autologous tumor cells or cell membranes processed in vitro to express α-gal epitopes bind the natural anti-Gal at the vaccination site and thus are opsonized and targeted to APCs. This type of immunotherapy can be performed with any human cancer, since all cells have carbohydrate chains that can be enzymatically processed to express α-gal epitopes. Patients with tumors that do not express TAAs will not be assisted by this vaccine, or by any other active immunotherapy method, since they have no Ags that differentiate between normal and malignant cells. However, in patients with tumors expressing TAAs, this autologous tumor vaccine may function as adjuvant immunotherapy providing the immune system with an additional chance to be effectively exposed to the TAA peptides and thus to mount an immune response that may eradicate the tumor cells in micrometastases and in minimal residue disease.

Abbreviations

Ab

Antibody

Ag

Antigen

APC

Antigen-presenting cell

DC

Dendritic cell

FcγR

Fcγ receptor

α-gal epitope

Galα1-3Galβ1-4GlcNAc-R

α1,3GT

α-1,3-Galactosyltransferase

KO mice

Knockout mice for α1,3GT

OVA

Ovalbumin

SA

Sialic acid

TAA

Tumor-associated antigen