Chemical synthesis of vaccines

An anti-cancer jab?

von Prof. Dr Horst Kunz, Sebastian Hartmann, Björn Palitzsch

Vaccinations against disease: a blessing for humanity

The discovery made by Emil von Behring in 1890 – namely that bacteria in humans and animals trigger a powerful immune response in the form of antibodies (immunoglobulins) targeting the invaders – represents a phenomenal advance in the history of medicine. His work, which paved the way for diphtheria and tetanus immunisation, was honoured with the first Nobel Prize for Physiology and Medicine in 1901. The basic principle of immunotherapy against infectious diseases is that molecules (saccharides, polysaccharides, proteins) in the bacterial cell wall are identified as foreign by the mammal organism and are combated by the formation of antibodies targeting these foreign structures. The immune response against viral infections such as flu proved much harder to trigger, since the viruses utilise the biosynthetical apparatus in mammalian cells for synthesising their membrane molecules. Accordingly, the organism does not directly identify these membrane molecules as foreign. Diseased cells such as tumour cells carry endogenic molecules produced by the organism itself (proteins, glycoproteins and glycolipids) on their membranes. These molecule structures are only weakly immunogenic: they are therefore shown natural tolerance by the organism’s immune system.

Over 30 years ago, however, G. F. Springer and co-workers [1] established that membrane glycoproteins in normal epithelial cells differ markedly from those found in epithelial tumour cells. The differences are not so much to do with protein sequences, but are to be found in the carbohydrate side chains. Biochemical research over the last 20 years has discovered that these structural differences are particularly applicable to the MUC1 mucin (episialin) and its tumour-associated form.

The MUC1 mucin as a target structure for an immune differentiation of normal and tumour cells

MUC1 is a glycoprotein anchored in the ­external cell membrane; it is highly glycosylated and extends far into the extracellular space (>100nm). It is expressed in almost all epithelial tissue (breast, prostate, colon, pancreas); it is generally strongly overexpressed in the equivalent tumour tissue. MUC1’s large extracellular portion contains an extended domain consisting of numerous repeat sequences – known as ‘tandem repeats’ – which vary in number according to the individual (Figure 1). Each of these repeat units incorporates five potential glycosylation sites – three threonines and two serines.

Fig. 1 MUC1 membrane glycoproteins on normal epithelial cells and epithelial tumour cells. Amino acids in single-letter coding.

In MUC1 as present on normal epithelial cells, the carbohydrate side chains are very long: they cover the protein backbone in this portion completely. The MUC1 molecule thus assumes an elongated form that protrudes far into its environment: it looks like a test tube brush thickly set with bristles. On account of the characteristically altered activity of glycosyltransferases ­(enzymes that transfer carbohydrates), the ­carbohydrate side chains in tumour cells are often severely shortened and prematurely sialylated, i.e. they terminate in the C-9 carbohydrate N-acetylneuraminic acid frequently found in mucus glycoproteins. These short saccharide side chains, alpha-glycosidically linked to threonine or serine, are held to be tumour-associated carbohydrate antigens [1–3]. We believe these short saccharide side chains make peptide sequences in the tandem repeat region of the protein backbone of tumour-associated MUC1 accessible to the immune system; in normal, healthy cells, these sequences are completely covered by the long saccharide side chains. Alongside the tumour-associated carbohydrate antigens (Figure 2), these sequences also then embody peptide structural information typical to tumours.

Fig. 2 Schematic display of the biosynthesis of mucin carbohydrate side chains. The five saccharide structures shown in the top part of the image are – bound to threonine or serine – tumour-associated carbohydrate antigens on epithelial cells. The corresponding fluorenylmethoxycarbonyl (Fmoc)-protected glycosyl amino acids (serine, threonine derivatives) are required for the synthesis of the tumour-associated glycopeptide antigens.

The shortened saccharide side chains in tumour-associated MUC1 clearly also have the effect of changing the conformations of the peptide chain. The elongated form can now develop knobbed folds (turn conformations) [4] that represent further structures identified as typical for tumours. From the structural differences discussed so far, one could conclude that the immunisation of patients with MUC1 mucin isolated from tumour cell membranes is able to induce a tumour-specific immune response. While antibodies of this kind are indeed found in cancer patient serum, they do not cause an immune response against tumour tissue because of the tolerance already mentioned. Furthermore, attempts to deploy tumour cell MUC1 for immunisation have failed largely due to the fact that every protein strand of tumour-associated MUC1 is populated both by short, tumour-associated saccharide antigens and by the long carbohydrate side chains typical for normal MUC1, in alternating combinations. The pure glycoprotein structures typical for tumours cannot be isolated. Modern chemical synthesis techniques can produce pure partial glycopeptide structures typical for tumours from MUC1, however [5]: since these are too weakly immunogenic, they must be linked to immune stimulating factors to produce effective vaccines.

MUC1 glycopeptide vaccines via linking with T helper cell epitopes

Considering the differences in molecular structure for the MUC1 mucin in normal epithelial cells on the one hand and tumour cells on the other, as outlined above, target structures for the immune system are set up in the form of glycopeptide antigens, in which the tumour-associated saccharide antigens – such as the TN antigen, the Thomsen-Friedenreich (T) antigen or their N-acetylneuraminic (sialic acid)-substituted forms, such as the sialyl-TN antigen – are linked with peptide sequences from the MUC1 tandem repeat region. Multi-stage chemical and stereoselective synthesis is required to obtain the chemically pure forms of the fluorenylmethoxycarbonyl (Fmoc)-protected glycosyl amino acid components. Compared to normal Fmoc amino acids, the glycosidic bonds present in these compounds (and in the products obtained from them) have the effect of producing an elevated sensitivity to bases and acids. They can be successfully deployed in the automated solid-phase synthesis of MUC1 glycopeptide antigens, however, if conditions are maintained as established by recent work in this field. The example shown in Figure 3 summarises the use of a 2-(trimethylsilyl)ethyl ester anchor (1) and a sialyl-TN threonine component [6].

Fig. 3 Solid-phase synthesis of a glycopeptide 2 from the tandem repeat region of the tumour-associated MUC1 mucin.

Cleavage of the completely protected glycopeptide was achieved by cleaving the anchor – in this case via fluoride-induced elimination under neutral conditions. Following hydrogenolysis of the benzyl ester, all acid-labile protective groups were removed from the amino acid side chains by using trifluoroacetic acid (TFA), triisopropylsilane (TIPS – serves to scavenge tert-butyl cations) and water. Particular attention is due to the cleavage of the O-acetyl groups, since the O-glycosidic bonds to serine/threonine are characteristically base-labile. A pH value of 11.4 should not be exceeded. As a rule, the glycopeptides (such as 2) were isolated in their pure form by subsequent preparative reversed-phase HPLC in quantities >10mg. Their purity and structure are not only verified by means of retention times in analytical HPLC and mass spectra but also via high-field NMR spectra [6].

Following a similar procedure in chemical synthesis, it is also possible to construct fully synthetic vaccines, in which the MUC1 glycopeptide is linked as a B cell epitope via an immunologically inactive oligoethylene glycol spacer with a T cell epitope peptide immune system stimulant, taken in this case from ovalbumin (Structure 3, Figure 4). Following purification via HPLC, this pure, wholly chemically synthesised vaccine was obtained in a quantity of 15mg – an adequate volume for immunising a veritable host of mice.

The subsequent presentation of the T cell peptide by the major histocompatibility complex (MHC) II on the B cell then effects – due to identification by the aforementioned CD4 receptor – the activation of the T helper cells, which ensure, via cytokines and co-stimulating factors, that this selfsame B cell transforms itself into a plasma cell and proliferates. This in turn triggers secretion of large quantities of antibodies, which are quasi equivalent to being copies of the immunoglobulin receptor on the B cell (Figure 4).

Fig. 4 Humoral immune response to a MUC1 glycopeptide-OVA T cell peptide conjugate 3, stimulated by TH cells.

Corroborating this, strong immune responses were triggered in the transgenic mice with synthetic vaccine 3. For their characterisation in an enzyme-linked immunosorbent assay (ELISA, Figure 5), one needs microtitre plates with immobilisable forms of the glycopeptide antigens against which the immune response is targeted. These can be obtained by conjugating these glycopeptides (such as 2) to carrier proteins. For linking to carrier proteins such as BSA, the glycopeptides were N-terminally extended in solid-phase synthesis by a triethylene glycol spacer amino acid. The fully unblocked form was transformed with bi-functional linker molecules – in the present scenario via squaric acid diesters at pH 8 – into the monoamide. The monofunctional form reacts with the -amino functions of the lysines in the protein in water (here, the squaric acid monoamide ester at pH 9.5) to yield the BSA conjugate. MALDI mass spectrometry shows it carries an average of six tumour-associated glycopeptide antigens in an unaffected structure [6]. This glycopeptide BSA conjugate is applied in an aqueous solution to the microtitre plate wells, where it remains adhered to the hydrophobic surface. Following rinsing with water one obtains a coating that one can understand as the simulation of a cell surface (Figure 5).

Fig. 5 ELISA test of mouse antibody (red) induced by vaccine 3, identified with sheep anti-mouse antibodies (bio = biotin, STA = streptavidin, HRP = horseradish peroxidase, ABTS = colourless substance, which forms a green radical cation via oxidation).

If one now applies the antiserum induced in the transgenic mouse by vaccine 3 to the microtitre plate, subsequent washing will leave only those mouse antibodies adhered to this conjugate that specifically target the glycopeptide (contained in the vaccine). These induced mouse antibodies are detected using a biotinylated sheep anti-mouse antibody, whose adhesion is, in turn, visualised using a streptavidin-horseradish peroxidase (HPO) conjugate, which catalyses the oxidation of a colourless heterocycle via hydrogen peroxide to a green radical cation. In the ELISA diagram (Figure 6), its absorption is plotted against the increasing dilution of the mouse antiserum (green line).

Fig. 6 ELISA identification of the immune response induced in transgenic mice by the synthetic vaccine 3 Black is negative control; Neutralisation of the mouse antiserum by glycopeptide 2

We found that such a strongly elevated immune response occurred in one of the three transgenic mice after the second booster dose with the fully synthetic vaccine 3. This is a marker for the formation of immunological memory and indicates a changeover from IgM to IgG type antibodies. Neutralisation of the antibodies induced in the mouse with the synthetic tumour-associated glycopeptide antigen 2 (red line) impressively demonstrates the high structural specificity of the immune response triggered. While the sialyl-TN glycopeptide 2 present in the vaccine completely neutralises the antibodies (Figure 6, red line), neither the same-sequence unglycosylated MUC1 peptide nor the sialyl-TN glycopeptide with a peptide sequence from the MUC4 mucin effect a neutralisation of these antibodies. One may therefore state as fact that synthetic vaccine 3 has triggered an eminently structurally selective immune response, which is targeted specifically against the MUC1 glycopeptide structure typical for the epithelial tumour cells.

The selectivity of the immune response induced by the synthetic vaccine is astonishing and highly promising for the ultimate goal of being able to achieve active immunisation in the patient against the patient’s own cancer cells. On average, however, the vaccine produces this effect only in every third mouse. Moreover, the transgenic mouse model is a model that cannot be extrapolated to humans.

Anti-tumour vaccines from MUC1 glycopeptide antigens and tetanus toxoid

To obtain glycopeptide vaccines that are not only reliably highly immunogenic but which could also be deployed in humans, we proceeded to link the synthetic tumour-associated glycopeptides via the linkage elements of an amino acid spacer and squaric acid linkers with tetanus toxoid as a carrier protein [7].

Tetanus toxoid is a protein with a molecular weight of >150,000 and which contains numerous T cell epitopes. It is a strong immunostimulant and has already been deployed successfully in many vaccines – such as flu vaccine – for human immunisation.

Synthesised using the described method, a MUC1 tetanus toxoid (TTox) vaccine, described in ref. [7] was used to vaccinate ten wild-type BALB/c mice. All ten animals exhibited a very strong immune response. Following the second booster dose, the ELISA test detected an antibody titre of >500,000, i.e. following dilution of the antiserum of the vaccinated mouse by a factor of 500,000, 50% of the absorption of the green radical cation was still found using ELISA (see above). This immune response is so powerful that it would break down the natural tolerance mentioned earlier against endogenic structures in each and every case. Moreover, it is also a structurally selective immune response targeting the tumour-associated MUC1 glycopeptide. If one adds this glycopeptide without TTox and squaric acid portion to the antiserum, then the induced antibodies are neutralised and binding to the MUC1 glycopeptide-BSA conjugate on the microtitre plate can no longer be detected using the ELISA test. This selectivity of the strong immune response is absolutely essential: if the induced antibodies would also bind to glycopeptide structures on normal epithelial cells, this could have undesirable consequences.

Regrettably, the antibodies triggered by this first synthetic TTox vaccine exhibited only weak binding to cells from the MCF-7 breast cancer cell line in a flow cytometer. In this kind of flow cytometer analysis, the tumour cells are first incubated with 1:1000 diluted serum from a vaccine-immunised mouse before then being washed and the bound mouse antibodies labelled with a fluorescent labelled goat anti-mouse antibody. During the fluorescent-activated cell sorting (FACS) analysis in the flow cytometer, all cells passing through are counted by Laser beam scattering. For the cells identified by the antibodies – and therefore fluorescing – a separate count is maintained.

Fig. 7 Anti-tumour vaccine 4 from a MUC1 glycopeptide glycosylated on Ser17 and tetanus toxoid.

Since the antibodies triggered by the vaccine with the sialyl-TN side chain excited a strong immune response in the mice, but the induced antibodies demonstrated only weak binding to the MCF-7 breast cancer cells in the flow cytometer analysis, we then used the technique as described to synthesise a MUC1 glycopeptide TTox vaccine 4, in which the glycosylation position is located at a different site (here: serine 17; see Figure 7). This vaccine also triggered a very strong, tolerance-breaking immune response in all mice [8]. In the FACS analysis (Figure 8a, negative control), the antibodies thereby formed now exhibit complete binding to the breast cancer cells (Figure 8b).

Neutralisation of the antiserum with the glycopeptide contained in the vaccine (equivalent to 2 in vaccine 3) cancels out this tumour cell labelling by the antibodies present in the antiserum (Figure 8c) [8].

Fig. 8 Flow cytometer FACS analysis of the binding of antibodies induced by vaccine 4 to breast cancer cells of the MCF-7 cell line

The binding of the vaccine 4 induced antibodies to the membrane glycoproteins of the MCF-7 breast cancer cells is reflected by the labelling of tumour cells in tissue sections. Figure 9 presents tissue sections of mammary carcinomas from three female patients incubated with antiserum taken from a mouse immunised with synthetic vaccine 4 (Figures 9a–c) [8].

The mouse antibodies binding to the tumour cells were in turn rendered detectable using a dye-coupled secondary antibody (here: a goat antibody) [8]. One observes that the antibodies induced by vaccine 4 barely bind at all to early-stage tumour tissue (Figures 9a,). They bind clearly to tumour tissue at the median G2 stage (Figure 9b), however, while the tumour cells taken from the advanced-stage tumour (G3, Figure 9c) are fully labelled by the antibodies triggered by vaccine 4. This variable degree of antibody labelling is undoubtedly attributable to the progression of the tumour and the concomitant rise in the proportion of tumour-typical MUC1 glycoprotein structures on the breast cancer cell membranes.

Fig. 9 Tissue sections from mammary carcinomas at different stages of progression; a)–c) following treatment with antiserum from a mouse immunised with vaccine 4 (colouration results from enzyme-linked secondary goat antibody, colour from oxidation of 3-amino-9-ethyl-carbazol).

Investigations of the subtypes of the antibodies triggered by the synthetic glycopeptide TTox vaccine have shown [7, 8] that the vast majority of antibodies formed are of the IgG subtype (IgG1). In accordance with immunological mechanisms, the labelling of the tumour cells by these specific antibodies should initiate the degradation of tumour cells by the immune system. These results entitle us to express the hope that, one day, immunisation with chemically synthesised vaccines using the procedure described here will achieve active immunisation of patients against their own tumour tissue, thus opening up vast new horizons in the field of cancer therapy.

Literature at the authors
1] G. F. Springer, Science 1984, 224, 1198.
[2] J. Taylor-Papadimitriou, J. M. Burchell, D. W. Miles, M. Daziel, Biochim. Biophys. Acta 1999, 1455, 301; I. Brockhausen, Biochem. Soc. Trans. 2003, 31, 6318.
[3] F.-G. Hanisch, Biochem. Soc. Trans. 2005, 33, 705.
[4] J. D. Fontenot, S. V. Mariappan, P. Catasti, No. Domenech, O. J. Finn, G. Gupta, J. Biol. Struct. Dyn. 1995, 13, 245.
[5] T. Becker, S. Dziadek, S. Wittrock, H. Kunz, Curr. Cancer Drug Targets 2006, 6, 491; O. Seitz, H. Kunz, Angew. Chem. Int. Ed. 1995, 34, 803; H. Kunz, S. Birnbach, Angew. Chem. 1986, 25, 360.
[6] S. Dziadek, A. Hobel, E. Schmitt, H. Kunz, Angew. Chem. Int. Ed. 2005, 44, 7630; S. Dziadek, D. Kowalczyk, H. Kunz, Angew. Chem. Int. Ed. 2005, 44, 7624.
[7] A. Kaiser, N. Gaidzik, U. Westerlind, D. Kowalczyk, A. Hobel, E. Schmitt, H. Kunz, Angew. Chem. Int. Ed. 2009, 48, 7551.
[8] N. Gaidzik, A. Kaiser, D. Kowalczyk, U. Westerlind, B. Gerlitzki, H. P. Sinn, E. Schmitt, H. Kunz, Angew. Chem. Int. Ed. 2011, 50, 9977.

Photo: © panthermedia | Sebastian Kaulitzki