Designing peptide-based FSL constructs to create Miltenberger kodecytes


  • 3A-S1-03

Steve Henry, Biotechnology Research Institute, AUT University, Auckland 1010, New Zealand


Background  KODE technology involves the synthesis of function-spacer-lipid (FSL) constructs and their insertion into cell membranes to create kodecytes. The functional group in FSLs includes blood group antigens and can be used to create synthetic antigen-modified red cells.

Aims  To review the issues in constructing peptide-based FSLs, with an emphasis on MNS system hybrid glycophorins and the resultant Miltenberger kodecytes.

Materials and Methods  Peptides suitable for synthesis into FSL constructs are ligated to a carboxymethylated oligoglycine spacer attached to a lipid tail. Simple contact of these FSL constructs with red cells allows for their spontaneous insertion into the cell membrane, creating kodecytes.

Results  Once optimized peptide sequences are identified, FSL constructs are simple and easy to build. Field trials established Miltenberger kodecytes were functional and able to identify clinically significant IgG antibodies.

Discussion  This review explores the issues with creating peptide-based FSL constructs, with an emphasis on the relatively simple MUT/Mur FSLs. With the recent development of easy-to-use construct-your-own FSL-peptide kits, researchers can now produce their own extended range of peptide-based blood group kodecytes.

Conclusion:  Provided possible complications/issues have been taken into account, FSL-peptides can be easily synthesized and used to make kodecytes representative of selected protein blood antigens/epitopes.


A lack of readily available rare or optimized/standardized red cell phenotypes has meant that alternative strategies are being devised to create ‘designer’ red cells with artificial or induced antigenic expression. Present methods capable of creating designer red blood cells with new antigenic profiles include transducing cord cells undergoing erythroid differentiation with viral vectors containing blood group alleles [1] or by attachment of antigens onto cells using KODE technology function-spacer-lipid (FSL) constructs. FSL constructs have been used in modifying embryos, spermatozoa, zebrafish, epithelial/endometrial cells, red blood cells, virions and to create quality controls systems, diagnostic panels, modify cell adhesion/interaction/separation/immobilization, and for in vitro and in vivo imaging of cells/virions [2–11].

This article will focus on the issues in designing peptide-based FSL constructs for creating kodecytes, and then specifically examine blood group MUT/Mur FSL constructs and the resultant vMNS kodecytes in current clinical use.

Kodecytes defined

Kodecytes are simply defined as cells that can be shown to have FSL constructs in their membrane [2–9], for example, MUT kodecytes [8]. Creating kodecytes is a simple and robust process with the most common procedure involving incubation of cells (in lipid-free media) with a solution of FSL constructs for a few hours at 37°C [2–9]. During incubation, the FSL construct(s) spontaneously incorporate into the cell membrane. Consistently reproducible results are obtained each time kodecytes are created provided strict control of incubation time, temperature, FSL concentration (including diluent formulation) and cell concentration are maintained. FSL constructs will remain in the red cell membrane (in vitro), if stored in lipid-free media, for the life of the cell.

FSL constructs

The creation of kodecytes requires FSL constructs. The basic structure (Fig. 1) of all FSL constructs is a functional head group (F), a spacer (S) and a lipid tail (L).

Figure 1.

 FSL construct created by ligating generic peptide via its cysteine side chain to the maleimide terminus of a partially carboxymethylated oligoglycine spacer coupled in turn via an adipic acid residue to the amino group of the phospholipid 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE).

Functional groups

The choice of the functional head group is primarily limited only by chemical and solubility issues. To date a large range of saccharides including blood group-related determinants, sialic acids, hyaluronin polysaccharides and other moieties such as fluorophores, biotin, reactive functional groups (RFGs) and a range of peptides have been used as the F groups on FSL constructs [2–11].


The spacer (S) is selected to provide a construct that is dispersible in water, yet will spontaneously and stably incorporate into a membrane. A variety of spacers have been constructed using appropriately functionalized oligomers of ethylene oxide, partially carboxylmethylated oligoglycines (of varying lengths and with/without additional modifications) and shorter building blocks like diaminoethane and adipic acid residues. The structure of a particular spacer reflects the needs of specific applications (including required distance from membrane and solubility demands) and the chemistry used for conjugation of the F group. In case the latter is represented by a plain peptide, highly selective and reliable thiol/maleimide ligation is employed. Notably, this approach imposes very modest restrictions on the functional peptide structure but implies incorporation of a thiol-bearing molecule (Fig. 1). While terminating the peptide chain with an extra cysteine residue is the most straightforward solution, simpler molecular relatives of cysteine such as cysteamine or 3-mercaptopropionic acid, may offer some advantage in particular cases. It should be emphasized that spacers suitable for use with undiluted human serum must be chemically unreactive and, ideally, be devoid of any detectable affinity towards serum and cellular components.


The lipid tail (L) is primarily for use in anchoring the construct to a surface, usually a cell membrane or to create micelle/liposomes. A range of lipids can be used (to impart specific membrane associations), but for red cells the lipid 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) is most suitable (Fig. 1).

Proteins, epitopes and antigenic peptides

The following is a brief overview of the related literature focused on issues relevant to our objective (for details on synthetic peptides antigenicity, the reader is referred to comprehensive reviews [12–21]). The term protein is usually used to describe a completely ordered or conformationally defined molecule, whereas peptide refers to a shorter oligopeptide chain amino acid oligomer (usually up to 20–30 amino acid residues). Both are comprised of amino acids covalently linked to each other via peptide bonds in a linear sequence (primary structure). Specific folding of the polypeptide chain results in a secondary structure with the most common features termed as alpha helix, beta sheets and turns. Tight packing of these secondary structural elements defines the stabilized tertiary structure of a protein, which may further associate with other proteins (or subunits) to form a protein complex (quaternary structure). Biologically active proteins (tertiary and quaternary structures) may also undergo conformational changes or allosteric shifts as they perform their function(s) or bind with antibodies [17–21]. Antigens (or epitopes – those parts of the antigen recognized by antibody) on proteins are divided into two types; linear or continuous epitopes and conformational or discontinuous epitopes [12–21]. Linear epitopes are continuous stretches of amino acids that can be easily imitated by the corresponding synthetic peptides. In contrast, conformational epitopes are formed by spatially separated short continuous elements brought together as a consequence of folding, and interactions within the longer polypeptide chain [12–21]. Clearly, the imitation of these epitopes with synthetic peptides is not straightforward and remains a formidable challenge.

Peptide mimotopes

Mimotopes offer an alternative approach to creating mimics of discontinuous epitopes, as they imitate epitope similarities in physicochemical properties and spatial organization, without a requirement of homology to the native antigen [22]. Mimotopes are selected purely on the basis of their interaction with the target, for example, antibody [23]. It is of note that peptide mimotopes can theoretically be made for almost any blood type, as exemplified by peptides representing carbohydrate blood group antigens [24,25]. However, mimotope peptides will still need to comply with FSL-specific limitations.

The primary issue for making a peptide-based FSL blood group determinant is finding an optimized peptide sequence based on the known antigen structure that will generate a soluble, chemically stable FSL construct correctly presenting the epitope at the RBC surface. Then the task is to find the balance between target affinity/specificity and cross-reactivity/non-specificity.

Antigenic peptide sequence (F) refinement in the context of FSL design

An initial step in finding a suitable segment of a protein for construction into an FSL is to speculate on those residues on the full protein sequence that are likely to be epitopes. This is generally very much easier for blood groups as the antigenic region of interest is usually tightly defined, in contrast to sequences on whole microorganisms where a large range of proteins and epitopes are potentially feasible diagnostic targets. One could expect for blood group polymorphisms a peptide representative of the polymorphic epitope by simply synthesizing a peptide that flanks variant amino acid residues. Unfortunately this simplistic approach is only sometimes successful, and usually only with linear epitopes. There is no doubt that the well-defined genetic mutation resulting in the amino acid change is responsible for the antigenic polymorphism, but the segment of amino acid residues containing the alternate amino acid is not necessarily the epitope. Substitution of amino acids that are remote from a particular epitope may lead to a conformational change in a discontinuous epitope [26–28] or effect the presentation of a linear epitope. Epitopes can also be more complex than just changes within a protein as interactions of one protein with another can create antigenic entities, for example, the Wright antigens [29].

Fortunately, some blood group antigens such as those on the glycophorins are essentially linear epitopes (e.g. MNS system) and are particularly suitable for engineering into FSLs.

Once candidate peptide sequences have been identified, they are analysed and refined to find those most likely to be compatible with, and successful as, an FSL construct. Some of these issues relate to the peptide [14,15] while others are specific to FSL chemistry. Many of the issues and characteristics of proteins/peptides/amino acids [12–21] can be identified by using existing algorithms/rules as found in resources such as: I-Tasser [30], BLAST [31], Immune Epitope Database (IEDB; [32], hydrophilicity [33], secondary structure [16], B-cell epitope [34], flexibility [35], surface accessibility [36], N-glycosylation [37], O-glycosylation [38] and epitope prediction [22].

Specific issues in the design and construction of peptide-based FSLs include those listed next (and for further technical details, the reader is referred to Technical Bulletins available at: These issues for consideration, where possible should be mitigated for, or compromises found. However, in many cases they have to be simply accepted as risks, with the outcome ultimately determined by success in both synthesis and biology.

Peptide length

In theory, the minimum number of amino acids required for antibody recognition is 4–6 [39]. But in practice anything less than 12 is not recommended as the additional flanking amino acid residues may also contribute to adequate epitope presentation and specificity, and increase the range recognition by polyclonal antibodies.

Internal cysteine

Peptide to spacer ligation chemistry (Fig. 1) employs reactivity of a thiol group introduced by coupling an extra cysteine residue to either the amino or carboxyl ends of the peptide (depending on orientation required) during its synthesis. In principle, this limitation requires that cysteine(s) be absent from the selected peptide sequence (although this limitation could be resolved either via isosteric substitution of Cys for α-aminobutyric acid [40], or potentially by using the native cysteine side chain for SL ligation).

N-terminal glutamine

Peptides bearing N-terminal glutamine (Gln) are known to undergo notoriously fast cyclization to form pyroglutamic acid residues [41] with the cyclization rate being in the range ∼2–3%/h under physiological conditions [42]. This spontaneous transformation is likely to effect antigenicity and should be excluded where the predicted epitope begins with N-terminal Gln. Sacrificing it, or selecting peptide with an extra residue preceding Gln, will remove this potential complication.

Asparagine deamidation

Peptides incorporating asparagine (Asn) followed by a non-hydrophobic residue –Asn–Xaa- are prone to spontaneous deamidation via intermediate formation of five-membered aspartimide, giving rise to a mixture of related α- and β-aspartyl peptides [43] potentially having altered antigenic profiles. The rate of degradation depends largely upon the Xaa structure and is highest for Gly approaching ∼2%/h under physiological conditions [44]. This type of intrinsic peptide instability is incurable by sequence manipulation and should be considered seriously. At least Asn–Gly should be excluded by all means while the potential problems in case of other Xaa should be evaluated on the basis of published data obtained for related model peptides [44] and experimental outcomes.


Perhaps the most important practical aspect of peptide selection focuses on detection and troubleshooting solubility problems associated with a particular sequence as such, and the related FSL construct. Ideally, to ensure smooth and efficient ligation, both should possess solubility in ligation buffer (0·1 m 4-methylmorpholine formate in 30% isopropanol, pH = 6·5) on the order of few mg/mL. Speaking from experience, we recommend avoiding sequences marked with the following insolubility signatures:

  •  peptides either having net charge |Zpept| < 2 themselves, or yielding FSL constructs with net charge in that region, that is, |Zpept − 5| < 2; Zpept = Σ(R+K+N-terminus + 0·5H) − Σ(D+E+C-terminus).
  •  sequences harbouring clusters formed by non-charged residues that include three or more consecutive I, F, Y, W, L, V, T.

In cases when solubility improvement through sequence manipulation is impossible, cosolvents can be used to solubilize peptide prior to ligation: (i) trifluoroethanol (or hexafluoroisopropanol, or 2-methoxyethanol) – ligation buffer or pyridine (1:1, v/v); (ii) neat dimethylformamide or (iii) 6 m guanidinium hydrochloride (should be tested in this order) and may prove useful for dissolving those peptides at a concentration of few mg/mL. Dimethyl sulphoxide (DMSO) may promote disulphide formation and should not be used to dissolve Cys-containing peptides. The resultant FSL construct may be more soluble than the starting peptide and although forcing conditions (e.g. increasing pH with ammonium bicarbonate, neat DMSO and aqueous alcohols) can allow for reconstitution of poorly soluble FSL constructs, all FSL constructs we use must ultimately be dispersible in saline alone. This is a critical feature of FSLs for biological use as it allows modification of cells without affecting their vitality and functionality.

Potential glycosylation sites

Glycosylation is a common post-translational modification of eukaryotic proteins and influences folding, solubility, antigenicity, conformation, interactions and half-life of peptides [37,45]. Both N- and O- potential glycosylation need to be considered during the peptide selection process. The standard N-glycosylation sequence is N-X-S/T where X could be any amino acid other than proline [46]. There is no single O-glycosylation motif but it usually occurs with high content of serine, threonine and proline residues [37] and in combinations like TAPP, TVXP, S/TPXP, TSAP, PSP and PST, where X is any amino acid [46]. As glycosylation dramatically changes the nature of an epitope the following strategies should be considered: (i) optimizing the peptide to avoid residues suspected in glycosylation or (ii) direct chemical synthesis of appropriately glycosylated peptides [47] or as a last resort (iii) using a naked peptide in the hope that it may still retain the expected binding specificity.

Needless to say, these considerations also apply to any post-translational modification [48,49] in the proximity of the epitope sequence.

Microbial relatedness

Algorithms for comparing primary biological sequences with microbial sequences (e.g. BLAST [31], IEDB [32]) help predict and evaluate the degree of potential cross-reactivity associated with the proposed peptide. Avoidance of microbially related epitopes reduces the risk of undesired non-specific cross-reactivity from naturally occurring antibodies directed against microbes.


No sequence is likely to be free from all the aforesaid potential issues and so compromises have to be made. In general, the perfect peptide candidate should be selected from the regions of the parent protein that are known or predicted to be exposed (loops, turns and both termini). It should possess the following virtues: hydrophilic and moderately polar, with low probability of glycosylation and other post-translational modifications and not be problematic in chemical sense (i.e. devoid of spontaneously degrading residues and internal Cys). It should also be water-soluble and preferably exist in the freely solvated non-hindered random coil conformation accessible for binding to antibody. During the final stage of design, to ensure native chain orientation, the peptide terminus to which the extra Cys for spacer ligation is to be added, is selected (Fig. 1).

In practice since the reliability of existing predictive methods of antigenicity are still unrefined, a series of at least four FSL constructs based on preferably unrelated candidate peptides should be evaluated.

‘Miltenberger’ FSLs and kodecytes

There is little doubt that IgG antibodies to hybrid MNS system glycophorins are clinically significant albeit still relatively poorly recognized and characterized [50]. In 2008, a series of ‘Miltenberger’ kodecytes bearing MUT and Mur were designed and constructed [8], based on peptides identified on the GP.Mur hybrid glycophorin (Fig. 2). Unlike their natural counterparts that always have multiple epitopes, for example, GP.Mur phenotype: Mia(MNS7), Mur(MNS10), Hil(MNS20), MINY(MNS34) and MUT (MNS35), kodecytes were created to express only individual Miltenberger epitopes. FSL construction analysis of MUT and Mur found no significant issues, except four sites of O-glycosylation and a moderate risk of deamidation for asparagine-26 in the MUT sequence (Fig. 2). Careful trimming of flanking residues with the aforesaid principles in mind resulted in successful preparation of water-dispersible FSL constructs whose structure, integrity and purity were unambiguously characterized (1H nuclear magnetic resonance spectroscopy and electrospray ionization mass spectrometry). A variety of FSL constructs were required to be made to optimize sensitivity and specificity [8]. A schematic diagram of the final optimized FSL-Mur construct is shown in Fig. 3.

Figure 2.

 Amino acid residues 20–50 of the GP-Mur hybrid glycophorin containing the MUT/Mur polymorphisms and the final FSL sequences considered most suitable for MUT/Mur kodecytes.

Figure 3.

 A schematically folded structure of the FSL construct based on the optimized Mur epitope, TYPAHTANE. Arrows indicate the beginning and end of the carboxymethylated oligoglycine-based spacer that separates the F and L segments.

Initial developmental testing [8] and later field testing [51,52] over the last few years have established that MUT kodecytes will detect both anti-MUT and anti-Mia while Mur kodecytes reliably detect anti-Mur. Interestingly both MUT and Mur kodecytes do not appear to effectively detect IgM activity, which although could be seen as a performance issue, is probably a desirable characteristic as IgM antibodies to antigens on hybrid MNS system glycophorins are not clinically significant [50]. The reason for this poor reactivity with IgM is uncertain but is potentially related to poor avidity/affinity of IgM to peptides [53,54]. Strategies to design multivalent FSL constructs may resolve the IgM reactivity issue – but unless IgG sensitivity is further enhanced it will probably not improve the diagnostic value of the existing monovalent FSL kodecytes. It is also important to note that monoclonal antibodies and FSL constructs only recognize and represent a portion of an antigen, respectively. This has created a tension between these two technologies, as unless they are aligned to the same peptide sequence, they will not react [8]. However, despite this FSL constructs for diagnostic antibody screening and identification are selected for their reactivity against clinically significant polyclonal antibodies.

With the recent development of new reactive functional group FSL construction kits for research use, researchers can now create their own FSL peptides at the laboratory bench. These kits require the user to only add peptide (with due consideration of the factors described before) and after drying they will have a fully functional FSL peptide of their own design. As a consequence, we expect to see in the near future a range of new and novel blood group-related FSLs.


The authors thank Tim Carroll, Stephen Parker, Debbie Blake, Alexander Tuzikov and Nicolai Bovin for their contributions to this work. Parts of the technical descriptions here are derived from KODE Biotech Materials Technical Bulletins and reproduced with permission.


SMH is a stockholder and the CEO/CSO of KODE Biotech the patent owner of KODE technology.