WO2002067860A2 - Selection by mirror image display - Google Patents

Abstract

Non-naturally occurring binders to cell surface carbohydrates and sugars are identified by a screening process that employs immobilized enantiomers of such cell surface carbohydrates and sugars. Preferred non-naturally occurring binders include D- peptides and L- nucleic acids and are resistance to enzymatic degradation and clearance. Single-chain Fab sequences that bind to sialic acid and KDO in nano-molar affinity were identified by this process. Exemplary screening procedures employed D-KDO, L-sialic acid and an L-sialo-disaccharide have been attached to a solid support for selection of high-affinity binders.

Description

SELECTION BY MIRROR IMAGE DISPLAY

Description

Field of Invention:

The invention related to processes for selecting, receptors for cellular surface carbohydrates. More particularly, the invention relates to solid supported mirror-image sugars and to their use for selecting receptors of biologically important carbohydrates.

Background: Carbohydrates on cell surfaces often form unique complex structures that act as key elements in various molecular recognition processes. Compounds that are capable of interfering with recognition of cell-surface carbohydrates thus hold great potential in biomedical applications. Such molecules, including inhibitors of carbohydrate-processing and synthesizing enzymes, cell-surface carbohydrate mimetics, as well as carbohydrate-binding antibodies, may be used as new drug candidates for the treatment of inflammation, cancer metastasis, bacterial or viral infection and as biosensors for detecting carbohydrates.

What is needed is effective means for identifying compounds that are capable of interfering with recognition of cell-surface carbohydrates, particularly compounds that are resistant to clearance mechanisms.

Summary:

The invention is directed to processes and reagents employable for selecting D-peptides and/or L-nucleic acids (RNA or DNA) having binding activity with respect to naturally occurring sugars or carbohydrates. The resultant D- peptide and/or L-nucleic acid binders are resistant to enzymatic degradation due to their non-naturally occurring chirality and are useful for blocking or activating the biological function of the naturally occurring sugar or carbohydrate to which they are targeted. Preferred naturally occurring sugars and carbohydrates are involved in disease processes. Examples include bacterial or viral cell-surface sugars or carbohydrates, heparan sulfates involved in viral entry, thrombosis, and angiogenesis.

The selection process employs the enantiomer, i.e., the mirror image compound, of the targeted naturally occurring sugar or carbohydrate. The enantiomer is synthesized or otherwise obtained and is then immobilized by attachment onto a support. The immobilized enantiomer is then employed to screen libraries of L-peptides and/or D-nucleic acids to identify active binders. Preferred libraries are phage display libraries. In an optional embodiment of the process, binders may be mutagenized or otherwise modified and rescreened with respect to the immobilized enantiomer so as to obtain and identify binders having enhanced binding activity. This process of enhancement may be reiterated as desired or until no further enhancement is achieved. The resultant L-peptides and/or D-nucleic acids having binding activity with respect to the target enantiomer are then isolated and identified. Enantiomers of the identified L- peptides and/or D-nucleic acids, i.e., D-peptides and/or L-nucleic acids corresponding to the identified L-peptides and/or D-nucleic acids, are then synthesized and tested for binding activity to the target naturally occurring sugar or carbohydrate. The synthetic D-peptides and/or L-nucleic acids that bind to the target naturally occurring sugar or carbohydrate may then be employed for blocking or enhancing the biological activity of these targets.

One aspect of the invention is directed to a process for selecting an L- peptide binder that binds to an enantiomer of a naturally occurring sugar or carbohydrate. In the first step of the process, the enantiomer of the naturally occurring sugar or carbohydrate is provided in a form employable for screening a library. Then, in the second step, the library is screened against the enantiomer for identifying the L-peptide binder having binding activity with respect to the enantiomer. The process may be extended by comparing the binding activity of the L-peptide binder both with respect to the enantiomer and with respect to the naturally occurring sugar or carbohydrate for determining the chiral specificity of the L-peptide binder identified above. The process may be further extended synthesizing the enantiomer of the L-peptide binder identified above for obtaining a D-peptide having binding activity with respect to the naturally occurring sugar or carbohydrate. And finally, the process may be further extended by contacting the naturally occurring sugar or carbohydrate with the D-peptide, under binding conditions, for blocking or enhancing the biological activity of the naturally occurring sugar or carbohydrate. Another aspect of the invention is directed to a process for obtaining a D- peptide having binding activity with respect to a first carbohydrate. The process employs first step of providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being employable for screening a library. Then, the library is screened against the second carbohydrate for identifying an L-peptide having binding activity with respect to the second carbohydrate. Then, the D-peptide which corresponding to the L-peptide is provided, the D-peptide having binding activity with respect to the first carbohydrate. In a preferred mode, this process may be extended for binding a D-peptide to a first carbohydrate. In this mode, the D-peptide is contacted with the first carbohydrate under binding conditions for binding the D-peptide to the first carbohydrate.

Another aspect of the invention is directed to a process for selecting an D- nucleic acid binder that binds to an enantiomer of a naturally occurring sugar or carbohydrate. In the first step of the process, the enantiomer of the naturally occurring sugar or carbohydrate is provided in a form employable for screening a library. Then, in the second step, the library is screened against the enantiomer for identifying the D-nucleic acid binder having binding activity with respect to the enantiomer. The process may be extended by comparing the binding activity of the D-nucleic acid binder both with respect to the enantiomer and with respect to the naturally occurring sugar or carbohydrate for determining the chiral specificity of the D-nucleic acid binder identified above. The process may be further extended synthesizing the enantiomer of the D-nucleic acid binder identified above for obtaining a L-nucleic acid having binding activity with respect to the naturally occurring sugar or carbohydrate. And finally, the process may be further extended by contacting the naturally occurring sugar or carbohydrate with the L- nucleic acid, under binding conditions, for blocking or enhancing the biological activity of the naturally occurring sugar or carbohydrate.

Another aspect of the invention is directed to a process for obtaining a L- DNA having binding activity with respect to a first carbohydrate. The process employs first step of providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being employable for screening a library. Then, the library is screened against the second carbohydrate for identifying an D-DNA having binding activity with respect to the second carbohydrate. Then, the L-DNA which corresponding to the D-DNA is provided, the L-DNA having binding activity with respect to the first carbohydrate. In a preferred mode, this process may be extended for binding a L- DNA to a first carbohydrate. In this mode, the L-DNA is contacted with the first carbohydrate under binding conditions for binding the L-DNA to the first carbohydrate.

Another aspect of the invention is directed to a process for obtaining a L- RNA having binding activity with respect to a first carbohydrate. The process employs first step of providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being employable for screening a library. Then, the library is screened against the second carbohydrate for identifying an D-RNA having binding activity with respect to the second carbohydrate. Then, the L-RNA which corresponding to the D-RNA is provided, the L-RNA having binding activity with respect to the first carbohydrate. In a preferred mode, this process may be extended for binding a L- RNA to a first carbohydrate. In this mode, the L-RNA is contacted with the first carbohydrate under binding conditions for binding the L-RNA to the first carbohydrate.

Another aspect of the invention is directed to an immobilized enantiomer of a naturally occurring sugar or carbohydrate employable for use in a selection process.

Another aspect of the invention is directed to a process for synthesizing a carbohydrate product. The carbohydrate product is an enantiomer of a first carbohydrate. The first carbohydrate is of a type capable of being synthesized by a first enantiomeric aldol reaction when catalyzed by a first A/-acetylneuraminic acid aldolase having a first enantiomeric specificity for catalyzing the first enantiomeric aldol reaction. The process includes the first step of converting the first Λ/-acetylneuraminic acid aldolase to a second Λ/-acetylneuraminic acid aldolase by directed evolution. The second Λ/-acetylneuraminic acid aldolase has a second enantiomeric specificity for catalyzing a second enantiomeric aldol reaction. The second enantiomeric aldol reaction is employable for synthesizing the carbohydrate product. Then, the carbohydrate product is synthesized by catalyzing the second enantiomeric aldol reaction with the second N- acetylneuraminic acid aldolase.

Brief Description of Drawing

Figure 1 illustrates the design of D-peptides and L-nucleic acids that target cell-surface sugars.

Figure 2 illustrates the synthesis of L-NeuNAc (1 ) and the disaccharide L- NeuNAc-a-(2-3)-L-Gal (2).

Figure 3 illustrates a scheme showing the synthesis of aminoethylthiopropyl-L-sialic acid (14) from L-NeuNAc (1 ) in five steps. Figure 4 illustrates three separate reactions or schemes. Reaction a) shows the immobilization of aminoethylthiopropyl-L-sialic acid on different solid supports. Reaction b) shows the conjugation of aminoethylthiopropyl-L-sialic acid to BSA. Reaction c) shows the immobilization of aminoethylthiopropyl-L-KDO on different solid supports. Figure 5 illustrates the schematic representation of the phage display method to select for antibodies against sugar-BSA conjugates.

Figure 6 illustrates a table showing the protein sequences for selected single-chain human antibodies (scFv).

Figure 7 illustrates a table showing the dissociation constants (nM) for single chain human antibodies (scFv) K18, S11 , and S18.

Figure 8 illustrates the polyvalent interactions between carbohydrate conjugates and surface immobilized single-chain antibodies.

Figure 9 illustrates the synthesis of L-sialic acid-BSA conjugate.

Figure 10 illustrates the cycle used in the directed evolution of N- acetylneuraminic acid aldolase to catalyze enantiomeric aldol reactions.

Figure 11 illustrates the biopanning procedure for peptide binders using phage display. Detailed Description:

A novel process is disclosed herein in which specific carbohydrate receptors that are enantiomers of natural peptides or nucleic acids can be identified through in vitro phage selection of peptides (K. Johnsson, L. Ge, in M. Famulok, E.-L. Winnacker, C.-H. Wong (Eds.): Combinatorial Chemistry in Biology, Vol. 243, Springer, Berlin 1999, pp. 87-106) or "evolution" of DNA (Famulok, M. Curr. Opin. Struct. Biol. 1999, 9, 324-329; Hermann, T.; Patel, D. J. Science 2000, 287, 820-825). The synthesis of the mirror image (the L-form) of the naturally occurring D-configurated carbohydrates is first obtained. This L-sugar is then used to screen phage expressing a peptide library on the coat proteins to identify specific clones that bind to the L-sugar. The mirror image of the L-peptide identified (i.e. the corresponding D-peptide) is then chemically synthesized and the D-peptide should bind to the natural form of the target (i.e. naturally occurring D-sugar). Similarly, unnatural L-DNA can be created to target specific cell-surface carbohydrates. This approach provides a new technology platform to study sugar-protein and sugar-nucleic acid interaction. It also serves as an effective way to design high-affinity and hydrolase resistant molecules as artificial receptors that are capable of binding to the natural carbohydrates (Figure 1 ). The advantage of this method is that large libraries of peptides or DNA can be easily generated and there are good methods available for selection of high-affinity binders that can be further improved through iterative process. This mirror-image approach has been used in identification of D-peptides (Schumacher, T. N. M.; et al. Science 1996, 271, 1854-1857) and L-RNA (Klussmann, S.; et al. Nat Biotechnol 1996, 14, 1112-5) to target L-peptide and D-adenosine respectively. Using this approach to target cell-surface carbohydrates, the unnatural enantiomers of carbohydrates should be readily available. As preferred examples, we disclose herein syntheses for the enantiomers of commonly found cell-surface carbohydrates, including N-acetyl-L-neuraminic acid (L-NeuNAc) 1 , the disaccharide L-NeuNAc-( -2,3)-L-Gal 2, 3-deoxy-L-manno-2-octulosonic acid (L-KDO) 3 and 3-deoxy-L-glycerogalacto-2-octulosonic acid (L-KDN) 4 and describe a general strategy for immobilization of these compounds on various solid supports for in vitro selection studies. The selection of single chain antibodies that bind to D-NeuAc with nanomolar affinity using the phage display method is also disclosed, as exemplary of the screening process.

L-NeuNAc 1 may be synthesized by either the neuraminic acid aldolase catalyzed addition of pyruvate to Λ/-acetyl-L-mannosamine (L-ManNAc) 5

(Gautheronlenarvor, C; et al. J. Am. Chem. Soc. 1991 , 113, 7816-7818; Lin, C. H.; et al. J. Am. Chem. Soc. 1992, 114, 10138-10145) or by chemical condensation of L-ManNAc with di-f-butyl oxaloacetate (Kuhn, R.; Baschang, G. Justus Liebigs Ann. Chem. 1962, 659, 156-161 ). L-ManNAc 5 was prepared from L-glucose as shown in Figure 2a. Anomeric benzylation of L-glucose followed by benzylidenation gave the diol 6 as white crystals (Yoshimoto, K.; et al. Chem. Pharm. Bull. 1979, 27, 2661-2674). Treatment of 6 with triflic anhydride (Knapp, S.; et al. J. Org. Chem. 1990, 55, 5700-5710; Knapp, S.; et al. J. Carbohydr. Chem. 1991 , 10, 981-993) followed by S_N2 substitution with sodium azide (Knapp, S.; et al. J. Org. Chem. 1992, 57, 7328-7334; Probert, M. A.; et al. Tetrahedron Lett. 1997, 38, 5861-5864) provided the azide 7 in 72% yield with complete inversion of C-2. A subsequent one-step catalytic reduction of the azido, benzyl, and benzylidene groups was attempted but the reaction did not go to completion. Compound 7 was therefore reduced in a stepwise fashion. Reduction of the azido group by triphenylphosphine followed by acetylation provided the mannosamine derivative 8 in 88% yield after crystallization. Catalytic hydrogenation over palladium(ll) hydroxide then gave the desired L-ManNAc 5 in gram quantities with 33% overall yield for the seven steps from L-glucose. Enzymatic synthesis of L-NeuNAc 1 via the neuraminic acid aldolase catalyzed condensation of mannosamine 5 with pyruvic acid (Lin, C. H.; et al. J. Am. Chem. Soc. 1992, 114, 10138-10145) was too slow to be useful for preparative synthesis. L-NeuNAc 1 was synthesized in 28 % yield by chemical condensation of L-ManNAc 5 and the potassium salt of di-f-butyl oxalacetate 11 (Figure 2b). (Kuhn, R.; Baschang, G. Justus Liebigs Ann. Chem. 1962, 659, 156-161 ) Compound 1 was further linked to L-galactose to provide 2.

D-NeuNAc-(α-2,3)-D-Gal exists as a common moiety in typical tumor-associated glycosphingolipid antigens and the binding site of influenza virus hemagglutinin (Hakomori, S.; Zhang, Y. Chem.& Biol. 1997, 4, 97-104; Ito, Y.; et al. Pure Appl. Chem. 1993, 65, 753-762). The enantiomer 2 was synthesized by coupling of L-sialyl phosphite 9 with 6-TBDPS-1-O-allyl-L-galactose 10, which was prepared from L-galactose through 1-O-allyl-2,3,4,6-tetraacetate-L-galactose 12 as shown in Figure 2c. The L-sialyl phosphite 9 was synthesized from L-sialic acid 1 by methylation of the carboxylic acid, acetylation of all the free hydroxyl groups, exchange the anomeric acetate to phosphite. Coupling of 9 and 10 was accomplished using TMSOTf at -78 °C to give the protected disaccharide 13 in 40% yield (Figure 2d) (Martin, T. J.; Schmidt, R. R. Tetrahedron Lett. 1992, 33, 6123-6126). Deprotection with TBAF followed by NaOH gave 2 in 80% yield. The anomeric configuration of the sialic acid in disaccharide 2 was determined to be α exclusively based on EXCIDE (Krishnamurthy, V. V. J. Mag. Res. Series A 1996, 121, 33-41 ; Ercegovic, T.; Magnusson, G. J. Chem. Soc, Chem. Commun. 1994, 831-832) (³J_C(1)._H(3)ax = 5.95 Hz).

D-KDO and D-KDN are key components of the core region of bacterial lipopolysaccharides (Hansson, J.; Oscarson, S. Curr. Org. Chem. 2000, 4, 535- 564). The enantiomers L-KDO 3 and L-KDN 4 were prepared on gram scales by neuraminic acid aldolase-catalyzed addition of pyruvate to L-arabinose or L-mannose, respectively, according to modified literature procedures (Gautheronlenarvor, C; et al. J. Am. Chem. Soc. 1991 , 113, 7816-7818; Lin, C. H.; et al. J. Am. Chem. Soc. 1992, 114, 10138-10145). We found that an increased amount of the enzyme was necessary to achieve high yield.

To immobilize the L-saccharides to solid support, 3-(2-aminoethyl-thio)-propyl moiety was chosen as the linker. It can be readily synthesized by irradiating allyl glycosides in the presence of cysteamine. Aminoethylthiopropyl-L-sialic acid 14 was prepared in five steps from L-NeuNAc 1 (Figure 3) (Roy, R.; Laferriere, C. A. Carbohydr. Res. 1988, 177, d-c4; Roy, R.; Laferriere, C. A. Can. J. Chem. 1990, 68, 2045-2054; Laferriere, C. A.; et al. Meth. Enzymol. 1994, 242, 271-280). Methylation and acetyl chloride treatment gave the fully protected sialyl chloride 15 in 94% yield. Allylation of 15 followed by basic hydrolysis provided the allyl sialoside 16 (50% yield), which was converted to 14 in 75% yield. It is necessary to keep the pH neutral to avoid decomposition of the compounds. Aminoethylthiopropyl-L-KDO 17 and L-KDN 18 were obtained in a similar manner. The aminoethylthiopropyl L-NeuNAc 14 and L-KDO 17 were immobilized through an amide bond on a number of solid supports such as Affi-Gel 15, Affi-Prep 10, Dynabeads M-270, and BIACore sensor chip CM5 (Figure 4a). In addition, we immobilized compounds 14 and 17 on Nunc immunoplates with two approaches. In the first approach we conjugated the compounds to bovine serum albumin (BSA, Figure 4b) and then absorbed the obtained conjugate on hydrophobic Nunc plates. In the second approach we directly immobilized compounds 14 and 17 on Nunc plates containing amino groups using the dicarboxylate linker or cyanuric chloride. To evaluate these novel carbohydrate conjugates the phage display technology was used (Scott, J. K.; Smith, G. P. Science 1990, 249, 386-390). We have previously shown that a human single-chain antibody phage library can be used to select for high-affinity human single-chain antibodies to tumor-associated carbohydrate antigens such as sialyl Lewis^x and Lewis" (Mao, S.; et al. Proc. Natl. Acad. Sci. USA 1999, 96, 6953-6958). We believed that the applicability of our immobilized carbohydrate entities for in vitro selection could be tested utilizing this antibody phage display library.

The library was subjected to four rounds of panning using Maxisorb Nunc plates coated with either D-KDO-BSA or D-NeuNAc-BSA (Figure 5). Phage were pooled after each successive round of panning and tested for their ability to bind either BSA conjugate. The enrichment clone number using either conjugate went from 2 x 10⁵ to 8 x 10⁸. Thus, after the fourth round of panning a total of 40 phage clones (20 to D-KDO-BSA and 20 D-NeuNAc-BSA) were randomly selected that survived the rigorous individual selection process. It was found that 14 and 16 clones exhibited substantial binding activity to

D-KDO-BSA and D-NeuNAc-BSA, respectively, as determined by ELISA. As a control, 20 random clones were collected from the unpanned library, and none of these clones showed any affinity (as determined by ELISA) to either D-KDO-BSA or D-NeuNAc-BSA. Clones with the greatest affinity as judged by ELISA were subjected to sequencing, and the nucleotide sequences encoding the V_H and V_L were determined. Of the 8 D-NeuNAc clones picked, two sequences were found that differed in the V_H and V_L regions, while of the 7 D-KDO clones chosen a single consensus sequence for both the V_H and V_L chains was uncovered (Figure 6). To analyze their binding specificity and affinity, the genes of the three novel single-chain antibodies (K18, S11 , and S18, see table) were subcloned into the pETFIag expression vector and purified from E. coll Analysis for the specificity and affinity of each single-chain antibody for their respective carbohydrate conjugates were determined by surface plasmon resonance (SPR) on BIAcore 2000. The single-chain antibodies were immobilized on the surface of the BIAcore CM5 chip. The carbohydrate conjugates were passed across the chip to test for antibody-carbohydrate interactions. Binding events measured as an increase in the SPR signal were monitored at several different analyte concentrations and fitted to a binding isotherm (see Experimental Section).

Figure 7 illustrates the specificity and affinity of the single-chain antibodies for the carbohydrate conjugates as determined by SPR. Since the antibodies were selected against D-sugars, they showed greater affinity for their respective D-sugar conjugates versus the L-sugar conjugates. The difference in dissociation constants is very significant between D-NeuNAc-BSA and L-NeuNAc-BSA. As expected, K18 displayed the greatest specificity for D-KDO-BSA as compared with L-KDO-BSA, since it was selected against D-KDO during the panning process. The antibodies selected against

D-NeuNAc, S11 and S18, displayed the greatest specificity for D-NeuNAc-BSA as opposed to L-NeuNAc-BSA, but showed no specificity between the D- and L-KDO-BSA. As a control experiment, no interaction between the antibodies and BSA that contains only conjugated linker but no sugar was observed. The mechanism of binding of the carbohydrate conjugates to the immobilized antibodies appears to be by polyvalent interactions (Figure 8). Unconjugated D-KDO (up to 50 mM) showed no interactions with any of the antibodies, and unconjugated D-NeuNAc showed binding at concentrations of 12.5 mM and above as judged by SPR (data not shown). When the carbohydrates were immobilized on the CM5 chip and the single-chain antibodies were passed over the chip surface, no significant interactions were observed in the micromolar range. Because each carbohydrate conjugate consists of 12 sugars linked to BSA (see Experimental Section), the observed nanomolar dissociation constants are probably a result of polyvalent interactions between the two interacting species.

This research report has presented our work toward the selection of biologically stable molecules capable of binding sugars and saccharides. A methodology for selecting the unnatural enantiomers of biopolymers that are capable of binding to the natural sugars and saccharides is outlined. Synthetic techniques for preparing the unnatural enantiomers of the sugars have been worked out. Preliminary studies using phage display libraries to select single chain antibodies against natural sugars have demonstrated the feasibility of this approach. Identification of short peptides capable of binding sugars and saccharides would enhance our understanding of sugar-protein interactions and provide a new approach to developing antibiotics and drug delivery systems. Selection of peptides against synthetic L-sugars is currently in progress in this laboratory.

Procedures for immobilization of L-saccharides onto polymer support. The aminoethylthiopropyl L-sialic acid and L-KDO were immobilized on a number of solid supports. Affi-Gel 15 (an agarose based resin) and Affi-Prep 10 (a methacrylate resin) are purchased from Bio-Rad as Λ/-hydroxysuccinimide activated form. The resin (1 mL) was shaken with a solution of aminoethylthiopropyl-sugar (20 mM) and triethylamine (5 μL) in anhydrous DMSO (500 μL) at 25 °C for 4 h. The excess carboxylate on the resin was blocked by incubating the resin with 500 μL of 1 M aqueous ethanolamine solution at pH 8.5 for 1 h. This afforded immobilization efficiency of about 90%. Dynabeads M-270 carboxylic acid magnetic beads (Dynal) were pre-activated by treatment with 0.1 M EDC and 0.1 M NHS solution in DMSO for 2h at 25 °C. The resin was shaken with a solution of aminoethylthiopropyl-sugar (20 mM) and triethylamine (30 mM) in anhydrous DMSO at 25 °C for 4 h. BIAcore sensor chip CM5 is supplied by the manufacturer in the free carboxylate form. It was activated by flushing with a solution of Λ/-hydroxysuccinimide and EDC for 7 min at 10 μL/min. Flushing with aminoethylthiopropyl-sugar in a pH 8.0 HEPES buffer containing 1 M sodium chloride (5 μL/min for 14 min) then provided the labeled chip. Ethanolamine solution (1 M, pH 8.5, 10 μL/min for 7 min) was then applied to quench the excess activated acid on the chip.

A 50 mM aminoethylthiopropyl-sugar and 60 mM DSS solution in anhydrous DMSO (0.1 mL) was incubated at 25 °C for 1 h. BSA (5 mg) solution in 0.1 M carbonate-bicarbonate buffer (0.9 mL) was added and incubated at 25 °C for 3 h. The BSA-sugar conjugate was purified using YM-10 Micron centrifuge filter devices (Millipore). We obtained about 3 mg of BSA conjugate with each sugar. MALDI-TOF analysis of the conjugates showed that on average 12 molecules of sugar are attached to each modified BSA molecule.

Determination of dissociation constants of single-chain antibodies on BIACORE 2000 instrument.

According to standard procedures, 4 μM of the single-chain antibodies were immobilized on the CM5 chip via the amino groups on the protein. Injections of the carbohydrate conjugates were pre-programmed into the BIACORE Control Software (version 1.3) and the concentrations ranged between 1.35 nM to 1350 nM to encompass the K_D value. At equilibrium conditions, 4 min. after each injection (10 μL/min), the response unit was recorded. The values were fitted to the following equation [AB]_8q = AB_max(1/(1 + MA])) to determine KD (Myszka, D. G. Meth. Enz. 2000, 323, 325-340).

Procedure for the Isolation of Peptide Binders of L-Sialic Acid via Phage Display

Preparation of 96 Well Plates

120μL of 100μg/mL of L-Sialic Acid conjugated BSA (Figure 9) in 50mM NaHepes/150mM NaCI (pH 7.4) was loaded into each well of the plate (high binding polystyrene plates Corning Costar #3690). The sugar conjugated BSA was allowed to bind via hydrophobic interactions to the polystyrene wall of the well for 3 days. The solution was pipetted off and 170μL of 5mg/mL of BSA in 50mM NaHepes/150mM NaCI (pH 7.4) was added to the wells to block any exposed binding sites on the polystyrene wall. This was done to minimize any nonspecific binding of the phage to the plate. At this point, a parallel set of wells was also blocked with the BSA solution. These wells were used for the control experiment. The wells were incubated with BSA for about 5 hrs. The BSA was then removed and the wells washed five times with 50mM

NaHepes/150mM NaCI (pH 7.4) to remove any residual BSA. The prepared plates were stored in 50mM NaHepes/150mM NaCI (pH 7.4) at 4 °C. Ten wells were prepared for five rounds of panning for each set of experiments, five with sugar conjugated BSA (for the binding experiment) and five with only BSA (for the control experiment).

Panning Procedure

(Figure 11 ) Phage and Ligand Incubation. Panning for L-Sialic Acid binders was done under low salt conditions.

Therefore, the wells were washed twice with 180μL of 50mM NaHepes (pH 7.4). For the first round of panning experiment, 20μL of the stock 12mer Phage Display Library (New England BioLabs) was diluted with 180μL of 50mM NaHepes (pH 7.4). 100μL of the phage solution was added to the well containing sugar conjuaged BSA. The other 100μL was added to the well containing only BSA. Incubation was allowed to occur overnight at 4 °C. Also, an overnight culture of 10mL of ER2738 E. coli (New England BioLab) was prepared in LB media and 15μg/mL tetracycline for the phage amplification step next day.

Remove Background and Elute Binders

The phage solution from each well was pipetted off and discarded. Each well was washed twice with 180μL of 50mM NaHepes/0.1% Tween (pH 7.4) to remove background binders. The bound phage was eluted by incubating 100μL of 5mM L-Sialic Acid in 50mM NaHepes (pH 7.4) for 2 hrs at 4 °C. 1 μL of eluted phage was used for titering (see below) while the rest was amplified in the next step. Phage Amplification and Isolation

Only the phage from the sugar binding experiment was amplified. The phage from the control experiment was discarded. While the elution process was taking place, 100mL of the cell culture was grown by a 1 :50 inoculation from the overnight culture. After 1 hr., the cell density was about 0.2 Abs. The eluted phage was added to the cell culture and the cells were allowed to grow for 5 hrs at 37 °C to amplify the bound phage. The cell culture was then centrifuged for 0.5 hr at 8,000 rpm to remove the cells. The phage, which resides in the supernatant, was precipitated with 20mL of 20% (w/v) PEG-8000/2.5M NaCI for 1 hr. at 4 °C. The phage was harvested by centrifugation at 2,000 rpm in 50 mL culture tubes at 4 °C. The amplified phage was resuspended in 400μL of 50mM NaHepes/150mM NaCI (pH 7.4) and stored at 4 °C. The first round of panning was accomplished.

For the second round of panning, the amplified phage was diluted with 50mM Na Hepes (pH 7.4) and the panning process begins again starting with the Phage and Ligand Incubation step. For rounds #2, 3, 4, and 5, background phage binders was removed with 10 wash cycles with 50mM NaHepes (pH 7.4)/0.1 % Tween. Five rounds of panning was done to isolate the L-Sialic Acid binder.

Phage Titering

10mL of LB/tetracycline (15μg/mL) was inoculated with an overnight culture of ER2738 and was allowed to grow until 0.6 Abs. While cells were growing, Agarose Top was melted and dispensed in 3mL sterile culture tubes and equilibrated at 45 °C until ready for use. Serial dilution of phage was prepared at 10^"2, 10^~4, and 10^"5 dilutions in sterile water. Once the cell culture was ready, 200μL of cells were dispensed into microfuge tubes, 1 for each dilution and 10μL of each dilution was added to the tube. The cells were incubated at room temperature for 5 minutes to allow for complete infection of the phage to occur. The infected cells were then transferred to the Agarose

Top, mixed, and poured onto pre-warmed LB/Tet/IPTG/Xgal plates. The plates were allowed to cool for 5 minutes, inverted, and incubated overnight at 37 °C. The titer was acquired only from blue/green plaques since these are made of cells infected with phage. After the fifth round of panning, the titer of the phage from the binding experiment, that containing L-Sialic Acid conjugated BSA, is then compared with the control experiment, that containing only BSA. If a binder exists, the titer from the binding experiment should be much greater than the titer for the control experiment, usually 5 to 10 times.

Purification and Sequencing of Phage

The overnight culture of ER2738 was diluted 1 :100 in LB media and dispensed in 2mL volumes into 15mL sterile culture tubes. With a sterile wooden stick, a stab of blue/green plaque was transferred to a culture tube containing the diluted cells. 10 plaques from the 5th round were picked and grown for 5 hrs at 37 °C. 1.2mL of the each culture was transferred to the microcentrifuge tube and centrifuged for 10 minutes. 1 mL of the phage containing supernatant was then transferred to a fresh microcentrifuge tube and 400μL of 20% PEG-8000/2.5M NaCI was added to it. The solution was mixed and let stand at room temperature for 10 minutes. The precipitated phage was then harvested by centrifugation for 10 minutes and the supernatant was discarded. The tube was centrifuged again and the remaining trace supernatant was pipetted off. The phage pellet was suspended in 100μL Iodide Buffer (1 OmM Tris-HCI/1 mM EDTA/4M Nal pH 8.0) and 250μL of ethanol was added to the tube. The solution was incubated for 10 minutes at room temperature and the single-stranded phage DNA was harvested by centrifugation for 10 minutes. The DNA pellet was washed with 70% ethanol to remove the remaining salts and dried under vacuum. The pellet was resuspened in distilled water and the DNA was sequenced using automated DNA sequencing methods.

Directed Evolution of Λ/-Acetylneuraminic Acid Aldolase to Catalyze Both D- and L- Enantiomeric Aldol Reactions Using error-prone PCR for in vitro directed evolution, the sialic acid aldolase from E. coli has been altered to improve its catalytic activity toward enantiomeric substrates including N-acetyl-L-mannosamine and L-arabinose to produce L-sialic acid and L-KDO, the mirror-image sugars of the corresponding naturally occurring D-sugars. The first generation variant containing two mutations (Tyr98His and Phe115Leu) outside the α/β-barrel active site exihibits an inversion of enantioselectivity toward KDO, and the second generation variant contains additional amino acid changes (Val251 lie) outside the α/β-barrel active site that significantly improves the enantiomeric reactions for L-sialic acid and L-KDO.

It is further disclosed by in vitro directed evolution that only a few mutations outside the active site of Λ/-acetylneuraminic acid aldolase (NeuδAc aldolase, EC 4.1.3.3) can significantly affect the enantioselectivity of this enzyme. The evolved enzyme is more effective as a catalyst for the synthesis of both D- and L-sugars, which are useful building blocks for development of mirror-image phage display technology. The work also provides useful system for understanding the factors that affect the enantioselectivity of enzyme catalysis. Λ/-acetylneuraminic acid aldolase catalyses the reversible aldol reaction of Λ/-acetyl-D-mannosamine and pyruvate to give Λ/-acetyl- D-neuranominic acid (D-sialic acid) and has been extensively used in the synthesis of sialic acids and analogs (Kim, M. J.; et al. J. Am. Chem. Soc. 1988, 110, 6481-6486; Lin, C.-C; Lin, C.-H.; Wong, C.-H. Tetrahedron Lett. 1997, 38, 2649-2652; Kragl, U.; et al. J. Chem. Soc, Perkin Trans. 1. 1994, 119-124; Machajewski, T. D.; Wong, C.-H. Angew. Chem., Int. Ed. 2000, 39, 1352-1374). With regard to its substrate specificity, NeuδAc aldolase has been shown to be specific for pyruvate as the donor, but flexible to a variety of hexoses and pentoses, both D- and to some extent L- sugars, as acceptor substrate (Lin, C. H.; et al. J. Am. Chem. Soc 1992, 114, 10138-10145; Gautheron-Le Narvor, C; et al. J. Am. Chem. Soc. 1991 , 113, 7816-7818). It is disclosed herein that the enantioselectivity of this enzyme has been engineered to develop an effective catalyst for the synthesis of both D- and L-sugars.

Directed Evolution of NeuδAc aldolase to cleave D- and L-KDO

A screening strategy similar to a previous approaches to KDPG aldolase was employed for this study (Fig. 10). The advantages of the screening approach are: (a) low background, (b) no need to consider the problem of cell permeation of substrate and (c) a similar screening system has already been established (Fong, S.; et al. Chem. Biol. 2000, 7, 873-883). The disadvantage is that it is time consuming. Coupling the aldol cleavage with the reduction of pyruvate to lactate with a concurrent oxidation of NADH to NAD in the presence of lactic dehydrogenase formed the basis of our screening methodology. Assay samples containing D- or L-KDO, the evolved crude enzyme, and the coupled system were prepared on 96-well plates and absorbance of NADH at 340 nm was monitored over time with a micro-titer plate reader. Screening of enzymes with low initial activities could be facilitated by the presence of a sufficient amount of enzyme in each sample, such that a good signal to noise ratio could be obtained over a reasonable period of assay time. The N-terminal 6x histidine fusion expression vector, pTrcHisB (Invitrogen) was chosen for our studies of NeuδAc aldolase as it provides a high expression system and the availability of rapid single step affinity purification of the expressed protein for further analysis. Subsequent libraries of mutant NeuδAc aldolase genes were cloned into this vector.

The wild-type aldolase gene was amplified under standard mutagenic PCR condition with a controlled mutation rate of 1-5 bases per gene to generate the first generation plasmid library. The protein library was prepared by lysis of re-suspended cell pellets harvested from 2mL cultures of individually picked colonies obtained by transformation with the plasmid library. The lysed suspensions were then heat treated to reduce background interference due to other contaminants present in the cell lysate, and the supernatants were transferred to 96-well plates for assays at 25 °C. About 25% of the first generation mutants still retained certain activity after the heat treatment. Hence, we adopted this strategy for reducing the background signal and for a more stringent evolution condition that restrains the drifting of thermal stability of selected mutants. One percent of the 1 ,600 members in the first generation were found more active toward D-KDO than the wild-type enzyme. Four mutants (0.25% screened population) amongst the first generation showed a significant enhancement in the rate of D-KDO cleavage. Unfortunately, no mutant with higher activity toward L-KDO was found. Compared with other studies, it is interesting that a relatively high fraction of population with beneficial mutations was observed in the first generation library, indicating that the enzyme may be poorly optimized toward the KDO cleavage reaction but is 5 a good starting template for the evolution of this property. The genes from four of the selected first generation mutants (Mut#1 , Mut#2, Mut#3 and Mut#4) that showed improvement in the cleavage of D-KDO were sequenced. It was found that Mut#1 , Mut#2, and Mut#3 had the same mutations. DNA shuffling was done to produce the second generation library using Mut#1 and Mut#4. Eight 0 hundred variants of this library were screened for both D- and L-KDO cleavage; however, no mutant with improved activity toward L-KDO was identified. Thus, the best first generation mutant (Mut#1 ) was selected as a template for another round of mutagenic PCR and screening. 1 ,600 variants of the error-prone PCR products were screened and one mutant (designated as Mut#2-5) with 5 cleavage activity toward L-KDO (0.06% screened population) was identified for further characterization. Characterization of the catalytic properties of the evolved NeuδAc aldolases

The selected mutants were purified and analyzed for their ability to cleave D- and L- KDO. Both the first generation mutant Mut#1 and the second 0 generation mutant Mut#2-5 have higher k^ and lower K_m toward D-KDO compared to the wild-type. Their specificity constant (k_∞t/K_m) toward D-KDO is 3.3- and 3.6-fold increase compared to the wild-type, respectively. Progressive improvement of both / _cat and K_m was observed in the evolution of the aldolase toward L-KDO cleavage. Compared with the wild-type, the second generation δ mutant Mut#2-δ has a 1.6-fold increase in k^ and a 1.δ-fold reduction in _m for L-KDO and hence a 2.4-fold improvement in k_∞υK_m. Interestingly, Mut#2-δ has a 30-fold increase in k^ toward L-sialic acid compared to the wild type, while the / _cat toward D-sialic acid remains basically the same. The K_m value is much higher for L-sialic acid, resulting in a relatively low / _cat, _m value. For synthetic 0 applications, however, the substrate concentration is often relatively high so that the enzyme operates at the maximum velocity. Similarly, Mut#1 and Mut#2-5 exhibit higher k^ toward D- and L-KDO compared to the wild type. With regard to the specificity constant measured by k^K_m Mut#2-5 showed slightly better L-enantioselectivity toward KDO (0.27 for L vs. 0.25 for D) while no inversion of enantioselectivity was observed for KDO or sialic acid. However, an inversion of enantioselectivity toward KDO was observed for Mut#1. Thus, while the wild-type enzyme has remarkable high substrate specificity for D-sialic acid over L-KDO, the evolved enzyme Mut#2-5 is significantly less specific for these substrates and an improved L-enantioselectivity was observed with KDO.

Condensation reaction of evolved aldolases The wild-type NeuδAc aldolase catalyzes reversible aldol reactions between pyruvate and sugar substrates. The enzyme is selective for sugar substrates with D-configuration , however it accepts several L-sugars at low reaction rate. We examined the effect of mutations on the addition reactions between pyruvate and D-, L-sugars respectively. Compared with wild-type, Mut#2-δ has 3.3-fold improvement in rate for addition o

Λ/-acetyl-L-mannosamine to pyruvate. Λ/-acetyl-L-mannosamine has been reported as a very weak substrate for the wild-type E. coli NeuδAc aldolase (Kim, M. J.; et al. J. Am. Chem. Soc 1988, 110, 6481-6486) (at -0.6% the rate for Λ/-acetyl-D-mannos-amine), but no detectable activity under the current assay condition. Though Mut#2-δ exhibits a relatively high k^for L-sialic acid, a relatively high concentration of substrate is required, however, as the affinity for the sugar substrate is relatively high. Further improvement in this regard is to lower the _m of Mut#2-δ for sialic acid and for N-acetyl-L-mannosamine. The activities of Mut#2-δ toward L-and D- arabinose are relatively high compared with wild-type and have been used in the preparative synthesis of D- and

L-KDO using the condition described previously (Gautheron-Le Narvor, C; et al. J. Am. Chem. Soc. 1991 , 113, 7816-7818).

Structural remodeling of mutations observed in the evolved aldolases The first generation mutants possess two common mutations, Tyr98His and Phe11 δLeu. An additional favorable cumulative mutation Val2δ1 lle for L-KDO was introduced in the second round of mutagenesis.and screening. According to the three-dimensional structure of this schiff base forming class I aldolase (Izard, T.; et al. Structure 1994, 2, 361-369) Phe11 δ is at the end of the loop between strand d and helix D. Tyr98 is located towards the C-terminal end of helix C. Val 2δ1 is located between helix I and helix J. None of these residues is in the active site. How do the changes of these amino acids affect the enzyme structure and enantioselectivity remains an interesting subject for investigation. To address this question, we have determined the x-ray crystal structure of Mut#2-δ and work is in progress to elucidate the molecular mechanism of the enantiomeric aldol reactions and determine the origin of enantioselectivity in the enzymatic catalysis. 0

General procedures

Nucleic acid manipulations were done according to standard procedure (Maniatis, T.; Fritsch, E. F.; Sambrook, J. Molecular Cloning: A Laboratory Manual, 1982). Restriction endonucleases and T4 DNA ligase were purchased δ from New England Biolabs. Tag polymerases were purchased from Stratagene. U V kinetic assays were performed on a Cary 3 Bio U V-vis spectrophotometer. DNA sequencing was performed on an ABI377A automated sequencer. The reactions were performed using thermal cycle sequencing conditions with fluorescent labeled terminators. Curve fittings were 0 done by non-linear least squares method using KaleidaGraph (Abelbeck

Software). 96-well plate samples were analyzed with a CERES UV900 Hdi reader (Bio-Tek instruments).

Chemical synthesis of Λ/-acetyl-L-mannosamine δ Λ/-Acetyl-L-mannosamine was prepared as reported previously (Kozlov,

I. A.; et al. ChemBioChem. 2001 , 2, 741-746). Starting from δg of L-glucose, 800 mg of Λ/-acetyl-L-mannosamine was obtained (Yield 13%).

Synthesis of 3-deoxy-D-manno-2-octulosonic acid (D-KDO) and 0 3-deoxy-L-manno-2-octulosonic acid (L-KDO)

D-KDO was synthesized by chemical condensation of D-arabinose and oxalacetate as described previously (Kozlov, I. A.; et al. ChemBioChem. 2001 , 2, 741-746). L-KDO was made by enzymatic synthesis (Kim, M. J.; et al. J. Am. Chem. Soc 1988, 110, 6481-6486; Kozlov, I. A.; et al. ChemBioChem. 2001 , 2, 741-746) using the wild-type NeuδAc aldolase expressed in E. coli. Both enantiomers of KDO were purified by AG-1 X-8 (HCO₃ ^") anion-exchange resin (Bio-rad) and re-crystallization from ethanol-water to remove pyuruvate δ contamination.

Plasmid Construction

The 0.9-kb E. coli NeuδAc aldolase gene was amplified by standard PCR from a genomic preparation of E. coli JM109, with primers N-NeuA 0 (δ'-ATC GCG GAT CCG ATG GCA ACG AAT TTA CGT G) (SEQ ID NO 33) and C-NeuA (δ'-ATC CGG AAT TCT CAC CCG CGC TCT TGC ATC) (SEQ ID NO 34) flanking the gene with restriction sites BamH I and EcoR I. The resulting fragment was purified, doubly digested with BamH I and EcoR I and was ligated into vector pTrcHisB (Invitrogen) digested with the same restriction δ enzymes. The ligation product was transformed into E. coli XL1 Blue-MRF' by electroporation (Dower, W. J.; et al. Nucleic Acids Res. 1988, 16, 6127-614δ). Plasmids recovered from transformants were screened by PCR for the presence of the aldolase insert. A positive plasmid clone Λ/eu>A-pTrcHisB was sequenced, used for protein expression and as mutation template for the 0 construction of the first generation library.

Mutagenic PCR was carried out under standard error-prone condition (Eckert, K. A.; Kunkel, T. A. Nucleic Acids Res. 1990, 18, 3739-3744). Primers N-NeuA and C-NeuA were used to amplify and mutate the template gene. Forty picomoles each of the primers were used and the reaction δ conditions were : 10 ng template /10 mM Tris-CI, pH8.3/δO mM KCI/1.δ mM MgCI₂/0.δ μl DMSO/0.2 mM MnCI₂/0.2 mM each of dATP, dTTP, dCTP, dGTP/2.δ units Tag polymerase (Stratagene), in a total volume of δO μL. The mixture was thermocycled for 30 rounds of 94 °C, 1 min; δδ °C, 1 min; 72 °C 1 min, and then 1 round of 72 °C for 2 mins. The library fragment was 0 gel-purified and cloned into pTrcHisB as described above and the resulting library construct was transformed into XL1 Blue-MRF', amplified and purified as plasmid miniprep (Qiagen). The presence of gene insert in the library constructs was confirmed by gel-electrophoresis with the parent vector pTrcHisB as reference. Heterogeneity of the first generation library construct was examined by transforming the plasmid into JM109 following by random picking of three colonies, plasmid extraction, and DNA sequencing. Each of the selected mutants contained one to five mutations. DNA shuffling was done according to the method of Stemmer (Stemmer,

W. P. Proc. Natl. Acad. Sci. USA 1994, 91, 10747-10761 ). Fragments of 60-100 bp were isolated and used for the reassembly. The substrates for shuffling were prepared by standard PCR of the two selected plasmids, Mut#1 and Mut#4, from the first generation library. A thermal cycling program of 9δ °C for 120 s; 94 °C for 30 s, δ5 °C for 30 s and 72 °C for 30 s (25 cycles) was used for reassemble PCR and the amplification of the reassembled product.

Library screening for KDO cleavage

Constructs harboring the mutant library were transformed into JM109 by electroporation. The transformed culture was spread on Luria-Bertani (LB) agar plates containing 50 μg/mLof ampicillin and incubated at 37 °C for 16 hours. Individual colonies were picked, replicated on a LB agar-ampicilin plate, and dispensed into 24-well plates that contained 2 ml of 2X YT/50 μg ml^"1 ampicillin/0.2 mM IPTG (the broth was shaken vigorously before dispensing into the plates). The plates were sealed and shaken at 37 °C/200 rpm in a shaker-incubator for 18 hours, centrifuged at 4000 rpm/4 °C for one hour, and the supernatant was carefully decanted. Each cell pellet was re-suspended in 0.4 ml of δO mM potassium phosphate buffer, pH 7.δ containing O.δ mg/mL of lysozyme. The plates were rapidly frozen in liquid nitrogen followed by thawing at room temperature, and then incubated at 65 °C for 20 minutes. Cell debris was collected by centrifugation at 4000 rpm, 4 °C for one hour. 80 μl of the supernatant from each well was transferred to a separate well on a 96-well plate. Upon incubation at 25 °C for at least 10 minutes, 25 μl of an assay solution containing δO mM potassium phosphate, NADH and L-lactic dehydrogenase at pH 7.δ was dispensed to each well with an

8-channel-repeating pipette. In each well, the starting concentration of NADH was 360 μM, and 0.1 U of lactic dehydrogenase was present. Baseline drift at 340 nm was monitored for two minutes. All samples had leveled or insignificant baseline drift. 2δ μl of D- or L-KDO at 18 mg/ml (13 mM starting concentration) in 60 mM potassium phosphate, pH 7.5 was added to each well and the absorbance at 340 nm was monitored at 20 second intervals continuously for 5 minutes after an initial 10 seconds of medium strength shaking. The activity of each mutant was reflected by the rate of decrease of O.D_340nm. Mutants that had unusually high activity were selected. The selected mutant colonies were picked from the replicated LB agar-ampicillin plate, grown overnight in LB-ampicillin medium, and their plasmids extracted as DNA minipreps (Qiagen) and sequenced. 0

Enzyme expression and purification

Selected plasmids were transformed into JM109. To express the protein, the starter culture was prepared by picking individual colonies and inoculated into 5 ml of LB-ampicillin medium, at 37 °C, 220 rpm overnight. The δ starter culture was added to δOO ml of LB-ampicillin medium, and was incubated at 37 °C, 220 rpm. Protein expression was induced at O.D._600nm = 0.4, by the addition of IPTG to a final concentration of 0.2 mM. Cells were harvested 6 hours after the induction, by centrifugation at 4 °C, 8000 rpm for 10 minutes and were stored at -78 °C. Cell pellet from 600 ml culture was 0 re-suspended in 20 ml of δO mM potassium phosphate pH 7.δ/δmM β-mercaptoethanol/300 mM NaCI, chilled on ice, and was lysed by passing through a French Press (SLM instruments, Urbana, IL) compressed to 1 ,600 Psi and then released to ambient pressure. The process was repeated three times. Cell debris was pelleted by centrifugation at 12,000g, 4 °C for 1 hour. 6 The supernatant was filtered through a 0.2 micron cellular acetate membrane filter (Corning), and was loaded onto a Ni²⁺-NTA-agarose column with a bed volume of 2.δml pre-equilibrated with the cell re-suspension buffer. The column was washed with 20 ml of buffer containing δO mM potassium phosphate pH 7.5/δmM β-mercaptoethanol/300 mM NaCI/δ% glycerol/10 mM 0 imidazole, and then 20 ml of buffer containing 60 mM potassium phosphate pH 7.δ/δmM β-mercaptoethanol. Bound enzyme was eluted with 60 mM potassium phosphate pH 7.δ/δmM β-mercaptoethanol/2δ0 mM imidazole, and was dialyzed extensively against δO mM potassium phosphate pH 7.δ/δmM β-mercaptoethanol at 4 °C. Eluted enzymes were analyzed with SDS-PAGE and were found to be >90% pure in all cases. Enzyme solution was frozen in liquid nitrogen and was stored at -78 °C prior to use. No activity lost was observed upon freezing and thawing the enzymes. Enzyme concentrations δ were determined by Bradford procedure (Bio-Rad) using BSA as calibration standard.

KDO cleavage assay

The activity was determined by the standard coupled assay with L-lactic 0 dehydrogenase (EC 1.1.1.27, Type II from rabbit muscle) and NADH. A typical assay was initiated by addition of an appropriate amount of D-KDO or L-KDO in δO mM potassium phosphate, pH 7.δ/δ mM β-mercaptoethanol, to a mixture of L-lactic dehydrogenase (0.8 U)/NADH (0.43 mM)/ aldolase (10-100 μg) in 60 mM potassium phosphate, pH 7.δ/δ mM β-mercaptoethanol. The total reaction δ volume was 800μl. Prior to the addition of the substrate, the mixture was pre-incubated at 26 °C for δ minutes. UV absorbance at 340 nm was recorded continuously for two minutes and the slope of the absorbance curve during the first 30 seconds was used for rate estimation.

0 Assay for addition reaction.

The aldol condensation activity was determined by the rate of depletion of pyruvate. Pyruvate concentration was determined by a method similar to that previously reported (Gautheron-Le Narvor, C; et al. J. Am. Chem. Soc 1991 , 113, 7816-7818). Reactions were initiated by the addition of the 6 aldolase to a mixture of pyruvate and sugars preincubated at 37 °C. Reactions were performed in 1 ml of 60 mM phosphate buffer, pH 7.6. Starting concentrations of pyruvate, sugars and enzyme were 10 mM, 260 mM and 30-300 μg/mL, respectively. 100 μl aliquots were withdrawn from the reaction mixture at different time points and quenched with 30 μl of 7% perchloric acid. 0 Samples were neutralized with 20 μl of 1 M NaOH. 160 μl of each neutralized sample was diluted to 1160 μl and assayed for pyruvate. Detailed Description of Figures

Figure 1 shows the design of D-peptides and L-nucleic acids that target cell-surface sugars. The enantiomer of a cell-surface sugar (L-sugar) is used as an affinity ligand for identification of L-peptides (from a phage display library) δ or D-DNA or D-RNA (via in vitro evolution) that bind the ligand. Through an iterative process, more tight-binding peptides or nucleic acids are selected and the corresponding enantiomers (i.e. D-peptides or L-DNA or L-RNA) are synthesized chemically to target the natural surface sugar.

Figure 2 shows the synthesis of L-NeuNAc (1) and the disaccharide L- 0 NeuNAc-a-(2→3)-L-Gal (2). Reagents and conditions: a) BnOH, TsOH; b)

PhCH(OMe)₂, camphor sulfonic acid, 66%; c) Tf₂0, pyridine, CH₂CI₂; d) NaN₃, DMF, 72%; e) H₂, Pd(OH)₂/C, MeOH, Ac₂0; f) PPh₃, THF followed by Ac₂0, MeOH, 88%; g) H₂, Pd(OH)₂/C, MeOH, AcOH, 100%; h)Me0H,NiCI₂/H₂0, 70%, 2h, 28% ; i) Ac₂0, NaOAc, reflux, 3h, 48%; j) BF₃ ^«Et₂0, allyl alcohol, CH₂CI₂, 0 δ °C, 2h, 60%; k) NaOMe, MeOH, rt, 1 h; I) TBDPSCI, DMF, imidazole, rt, 2h,

72%;m) MeOH, amberlite IR-120, 2h, rt; n) pyridine, Ac₂0, rt, 12h; o) BF₃ ^»Et₂0, p-thiocresol, CH₂CI₂, rt, 2h; p) NBS, acetone, water, rt, 2h; q) diethyl chlorophosphite, CH₂CI₂, /Pr₂EtN, 1 h, rt, 10%; r) 10, TMSOTf, CH₂CI₂, CH₃CN, -78 °C, 1 h, 40%; s) TBAF, THF, rt, 12h; t) NaOH, THF, rt, 1 h, 80%. 0 Figure 3 is a scheme showing the synthesis of aminoethylthiopropyl-L- sialic acid (14) from L-NeuNAc (1) in five steps. Reagents and conditions: a) MeOH, HCl, rt, 12h; b) AcCI, AcOH, MeOH, rt, 48h, 94%; c) allyl alcohol, silver(l) salicylate, 3 A molecular sieves, 2δ °C, 18 h; d) MeOH, NaOMe, rt, 1 h; e) NaOH, H₂0, rt, 1 h, 60%; f) HSCH₂CH₂NH₂, H₂0, UV, 18 h, 7δ%. δ Aminoethylthiopropyl-L-KDO (17) and aminoethylthiopropyl-L-KDN (18) were obtained in a similar manner.

Figure 4 shows three separate reactions or schemes. Reaction a) shows the immobilization of aminoethylthiopropyl-L-sialic acid on different solid supports. The aminoethylthiopropyl L-NeuNAc 14 and L-KDO 17 were 0 immobilized through an amide bond on a number of solid supports such as Affi-Gel 1δ, Affi-Prep 10, Dynabeads M-270, and BIACore sensor chip CMδ. Reaction b) shows the conjugation of aminoethylthiopropyl-L-sialic acid to BSA. In addition, compounds 14 and 17 were immobilized on Nunc immunoplates with two approaches. In the first approach the compounds were conjugated to bovine serum albumin (BSA, Scheme 3b) and then absorbed the obtained conjugate on hydrophobic Nunc plates. In the second approach compounds 14 and 17 were directly immobilized on Nunc plates containing amino groups using the dicarboxylate linker or cyanuric chloride. Reaction c) shows the immobilization of aminoethylthiopropyl-L-KDO on different solid supports.

Figure 5 illustrates the schematic representation of the phage display method to select for antibodies against sugar-BSA conjugates. The library was subjected to four rounds of panning using Maxisorb Nunc plates coated with either D-KDO-BSA or D-NeuNAc-BSA. Phage were pooled after each successive round of panning and tested for their ability to bind either BSA conjugate. The enrichment clone number using either conjugate went from 2 x 10⁵ to 8 x 10⁸. Thus, after the fourth round of panning a total of 40 phage clones (20 to D-KDO-BSA and 20 D-NeuNAc-BSA) were randomly selected that survived the rigorous individual selection process. It was found that 14 and 16 clones exhibited substantial binding activity- to D-KDO-BSA and D-NeuNAc-BSA, respectively, as determined by ELISA.

Figure 6 is a table showing the protein sequences for selected single-chain human antibodies (scFv). FR = Framework region, CDR = complementarity-determining region. Clone K18 was selected from D-KDO, clones S11 and S18 were selected from D-NeuNAc.

Figure 7 is a table showing the dissociation constants (nM) for single chain human antibodies (scFv) K18, S11 , and S18. The dissociation constants were measured by the method in which the single-chain antibodies were immobilized on the BIACORE CM5 chip surface and the carbohydrate-BSA conjugate was injected into the flow cell at different concentrations. K18 was selected from D-KDO. S11 and S18 were selected from D-NeuNAc.

Figure 8 illustrates the polyvalent interactions between carbohydrate conjugates and surface immobilized single-chain antibodies. The interaction was detected by surface plasmon resonance on the CMδ chip surface via binding experiments using BIACORE 2000. Unconjugated D-KDO (up to 50 mM) showed no interactions with any of the antibodies, and unconjugated D-NeuNAc showed binding at concentrations of 12.5 mM and above as judged by SPR (data not shown). When the carbohydrates were immobilized on the CMδ chip and the single-chain antibodies were passed over the chip surface, no significant interactions were observed in the micromolar range. Because each carbohydrate conjugate consists of 12 sugars linked to BSA (see δ Experimental Section), the observed nanomolar dissociation constants are probably a result of polyvalent interactions between the two interacting species.

Figure 9 shows the synthesis of L-sialic acid-BSA conjugate. The scheme shows that 12 sialic acids are attached to one BSA protein. Figure 10 shows the cycle used in the directed evolution of N- 0 acetylneuraminic acid aldolase to catalyze enantiomeric aldol reactions.

Figure 11 illustrates the biopanning procedure for peptide binders using phage display. For the first round of panning experiment, 20μL of the stock 12mer Phage Display Library (New England BioLabs) was diluted with 180μL of δOmM NaHepes (pH 7.4). 100μL of the phage solution was added to the 5 well containing sugar conjuaged BSA (Step l ). The phage solution from each well was pipetted off and discarded. Each well was washed twice with 180μL of δOmM NaHepes/0.1 % Tween (pH 7.4) to remove background binders (Step 2). The bound phage was eluted by incubating 100μL of δmM L-Sialic Acid in δOmM NaHepes (pH 7.4) for 2 hrs at 4 °C (Step 3). 1 μL of eluted phage was 0 used for titering while the rest was amplified in the next step. The eluted phage was added to the cell culture and the cells were allowed to grow for δ hrs at 37 °C to amplify the bound phage (Step 4). The cell culture was then centrifuged for O.δ hr at 8,000 rpm to remove the cells. The phage, which resides in the supernatant, was precipitated with 20mL of 20% (w/v) PEG-8000/2.5M NaCI for 5 1 hr. at 4 °C. The phage was harvested by centrifugation at 2,000 rpm in 50 mL culture tubes at 4 °C (Step 5). The amplified phage was resuspended in 400μL of 50mM NaHepes/150mM NaCI (pH 7.4) and stored at 4 °C. The first round of panning was accomplished.

Claims

What is claimed is:

1. A process for selecting an L-peptide binder that binds to an enantiomer of a naturally occurring sugar or carbohydrate, the process comprising the following steps:

Step A: providing the enantiomer of the naturally occurring sugar or carbohydrate, the enantiomer being employable for screening a library; and then Step B: screening the library against the enantiomer of said Step A for identifying the L-peptide binder having binding activity with respect to the enantiomer.

2. A process according to claim 1 further comprising the following additional step: Step C: comparing the binding activity of the L-peptide binder both with respect to the enantiomer and with respect to the naturally occurring sugar or carbohydrate for determining the chiral specificity of the L-peptide binder identified in said Step B.

3. A process according to claim 2 further comprising the following additional step: Step D: synthesizing the enantiomer of the L-peptide binder identified in said Step C for obtaining a D-peptide having binding activity with respect to the naturally occurring sugar or carbohydrate.

4. A process according to claim 3 further comprising the following additional step: Step E: contacting the naturally occurring sugar or carbohydrate with the D- peptide of Step D under binding conditions for blocking or enhancing the biological activity of the naturally occurring sugar or carbohydrate.

5. A process for selecting a D-nucleic acid binder that binds to an enantiomer of a naturally occurring sugar or carbohydrate, the process comprising the following steps:

Step A: providing the enantiomer of the naturally occurring sugar or carbohydrate, the enantiomer being employable for screening a library; and then Step B: screening the library against the enantiomer of said Step A for identifying the D-nucleic acid binder having binding activity with respect to the enantiomer. 0

6. A process according to claim δ further comprising the following additional step:

Step C: determining the chiral specificity of the D-nucleic acid binder identified in said Step B by comparing the binding activity of the D-nucleic acid δ binder both with respect to the enantiomer and with respect to the naturally occurring sugar or carbohydrate.

7. A process according to claim 6 further comprising the following additional step: 0 Step D: obtaining an L-nucleic acid having binding activity with respect to a naturally occurring sugar or carbohydrate by synthesizing the enantiomer of the D-nucleic acid binder identified in said Step C.

8. A process according to claim 7 further comprising the following additional 6 step:

Step E: contacting the naturally occurring sugar or carbohydrate with the L- nucleic acid binder of said Step D under binding conditions for blocking or enhancing the biological activity of a naturally occurring sugar or carbohydrate. 0

9. An immobilized enantiomer of a naturally occurring sugar or carbohydrate employable for use in a selection process.

10. A process for obtaining a D-peptide having binding activity with respect to a δ first carbohydrate, the process comprising the following steps:

Step A: providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being employable for screening a library; then Step B: screening the library against the second carbohydrate of said Step A 0 for identifying an L-peptide having binding activity with respect to the second carbohydrate; and then Step C: providing the D-peptide corresponding to the L-peptide identified in said Step B, the D-peptide having binding activity with respect to the first carbohydrate. 6

11. A process for binding a D-peptide to a first carbohydrate, the process comprising the following steps:

Step A: providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being 0 employable for screening a library; then

Step B: screening the library against the second carbohydrate of said Step A for identifying an L-peptide having binding activity with respect to the second carbohydrate; then Step C: providing the D-peptide corresponding to the L-peptide identified in 5 said Step B, the D-peptide having binding activity with respect to the first carbohydrate; and then Step D: contacting the D-peptide of said Step C to the first carbohydrate under binding conditions for binding the D-peptide to the first carbohydrate.

0

12. A process for obtaining an L-DNA having binding activity with respect to a first carbohydrate, the process comprising the following steps:

Step A: providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being 5 employable for screening a library; then

Step B: screening the library against the second carbohydrate of said Step A for identifying A D-DNA having binding activity with respect to the second carbohydrate; and then Step C: providing the L-DNA corresponding to the D-DNA identified in said 0 Step B, the L-DNA having binding activity with respect to the first carbohydrate.

13. A process for binding an L-DNA to a first carbohydrate, the process comprising the following steps: δ Step A: providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being employable for screening a library; then Step B: screening the library against the second carbohydrate of said Step A for identifying A D-DNA having binding activity with respect to the second 0 carbohydrate; then

Step C: providing the L-DNA corresponding to the D-DNA identified in said Step B, the L-DNA having binding activity with respect to the first carbohydrate; and then Step D: contacting the L-DNA of said Step C to the first carbohydrate under 5 binding conditions for binding the L-DNA to the first carbohydrate.

14. A process for obtaining an L-RNA having binding activity with respect to a first carbohydrate, the process comprising the following steps:

Step A: providing a second carbohydrate, the second carbohydrate being an 0 enantiomer of the first carbohydrate, the second carbohydrate being employable for screening a library; then Step B: screening the library against the second carbohydrate of said Step A for identifying a D-RNA having binding activity with respect to the second carbohydrate; and then

Step C: providing the L-RNA corresponding to the D-RNA identified in said Step B, the L-RNA having binding activity with respect to the first carbohydrate.

15. A process for binding an L-RNA to a first carbohydrate, the process comprising the following steps: Step A: providing a second carbohydrate, the second carbohydrate being an enantiomer of the first carbohydrate, the second carbohydrate being employable for screening a library; then Step B: screening the library against the second carbohydrate of said Step A for identifying a D-RNA having binding activity with respect to the second carbohydrate; then

Step C: providing the L-RNA corresponding to the D-RNA identified in said Step B, the L-RNA having binding activity with respect to the first carbohydrate; and then Step D: contacting the L-RNA of said Step C to the first carbohydrate under binding conditions for binding the L-RNA to the first carbohydrate.

16. A process for synthesizing a carbohydrate product, the carbohydrate product being an enantiomer of a first carbohydrate, the first carbohydrate being of a type capable of being synthesized by a first enantiomeric aldol reaction when catalyzed by a first Λ/-acetylneuraminic acid aldolase having a first enantiomeric specificity for catalyzing the first enantiomeric aldol reaction, the process comprising the following steps:

Step A: converting the first Λ/-acetylneuraminic acid aldolase to a second N- acetylneuraminic acid aldolase by directed evolution, the second N- acetylneuraminic acid aldolase having a second enantiomeric specificity for catalyzing a second enantiomeric aldol reaction, the second enantiomeric aldol reaction being employable for synthesizing the carbohydrate product; and then Step B: synthesizing the carbohydrate product by catalyzing the second enantiomeric aldol reaction with the second Λ/-acetylneuraminic acid aldolase.