2,5-Anhydro-d-Mannose End-Functionalized Chitin Oligomers Activated by Dioxyamines or Dihydrazides as Precursors of Diblock Oligosaccharides

Diblock oligosaccharides based on renewable resources allow for a range of new but, so far, little explored biomaterials. Coupling of blocks through their reducing ends ensures retention of many of their intrinsic properties that otherwise are perturbed in classical lateral modifications. Chitin is an abundant, biodegradable, bioactive, and self-assembling polysaccharide. However, most coupling protocols relevant for chitin blocks have shortcomings. Here we exploit the highly reactive 2,5-anhydro-d-mannose residue at the reducing end of chitin oligomers obtained by nitrous acid depolymerization. Subsequent activation by dihydrazides or dioxyamines provides precursors for chitin-based diblock oligosaccharides. These reactions are much faster than for other carbohydrates, and only acyclic imines (hydrazones or oximes) are formed (no cyclic N-glycosides). α-Picoline borane and cyanoborohydride are effective reductants of imines, but in contrast to most other carbohydrates, they are not selective for the imines in the present case. This could be circumvented by a simple two-step procedure. Attachment of a second block to hydrazide- or aminooxy-functionalized chitin oligomers turned out to be even faster than the attachment of the first block. The study provides simple protocols for the preparation of chitin-b-chitin and chitin-b-dextran diblock oligosaccharides without involving protection/deprotection strategies.


S1 Preparation and characterization of A n M oligomers
Chitin oligomers of the type A n M, where A n represent uninterrupted sequences of N-acetyl Dglucosamine (A) residues with a 2,5-anhydro-D-mannose (M) residue at the reducing end, were prepared by nitrous acid (HONO) depolymerisation of chitosan (F A = 0.48), using an excess HONO (1.3 equivalents) to the fraction of D-glucosamine (D) residues 1, 2 . The excess of HONO converts all the D residues into M residues, whereas A residues are unaffected, providing solely A n M oligomers.
The water-soluble oligomers (DP < 10) were fractionated by preparative gel filtration chromatography (GFC) to obtain purified low molecular weight A n M oligomers of specific DPs ( Figure S1a

S2 Conjugation of A n M oligomers to ADH and PDHA studied by time course NMR
Conjugation of A n M oligomers to ADH and PDHA was monitored by time course NMR as described earlier 5 . In brief, conjugation reactions were performed in deuterated NaAc-buffer (with TSP added as internal standard) at pH 3.0, 4.0 or 5.0 using 2 equivalents of ADH or PDHA relative to the concentration of oligomers (20.1 mM). 1 H-NMR spectra were recorded at specific time points ( Figure   S3 and S4) and yields of conjugates were determined by integration of the obtained spectra. Minor resonances close to the major resonances from the hydrazones/oximes (H1, M E or Z) were attributed to the conjugation of oligomers with alternative forms of the M residue (H1, M' E or Z) (assigned in Figure S3 and S4). Both the major and minor resonances from the E and Z hydrazones or oximes were integrated to obtain the yield of conjugates. Due to overlapping resonances (e.g. H2, M Z overlapping with H1, M (gem-diol) in Figure S4), yields could not be calculated from the sum of integrals from the H1 reducing end proton resonances in each individual spectrum. Due to the slight polydispersity of the oligomers, the yields could not be calculated from the integrals of the resonances resulting from the oligomers. Therefore, the integral of the resonance from the internal standard (TSP, 0 ppm) was used as reference to calculate yields. As the conjugates were reduced in a subsequent step (see below) the integral value of the TSP resonance was related to the integrals of the resonances from the completely reduced conjugates (for each individual reaction mixture). The TSP integral value was set to this specific value in all the spectra obtained during the time course NMR study for the conjugation and reduction to calculate the yields of conjugates and subsequently reduced secondary amine conjugates during the reactions. The yields were calculated based on the assumption of 100 % yield of reduced secondary amine conjugates. Integrals of overlapping resonances were calculated by subtraction.

S3 Kinetic modelling of the reductive amination reaction
The reductive amination of oligosaccharides with different amines (e.g. oxyamines (PDHA) or hydrazides (ADH)) is comprised of several individual reactions with independent rates and rate constants. The overall reaction involves the conjugation (amination) of the oligosaccharide, where Eand Z-oximes or hydrazones (Schiff bases) are formed for oxyamines or hydrazides, respectively. For oligosaccharides where the reducing end aldehyde is in equilibrium with a hemiacetal (normal reducing end), the acyclic Schiff bases are in equilibrium with cyclic N-glycosides (e.g. Npyranosides). By adding a reducing agent, the Schiff bases will be irreversibly reduced forming secondary amine conjugates. Irreversible reduction of oligosaccharides by the reducing agent will prevent the reductive amination reaction from going to completion. The general reaction scheme is shown in Figure S5. Figure S5: General reaction scheme for the reductive amination of oligosaccharides with normal reducing ends including assigned rate constants for each independent reaction involved. Reversible reactions are described by two rate constants (forward and reverse), whereas irreversible reactions are described by one rate constant (the scheme indicates the assumption that reduction of E and Z had the same rate constant (k 5 )).
When considering the reactions to be first order with respect to each reactant, reaction rates can be determined by the following equations chosen sufficiently small to result in a simulation which did not further change when choosing an even smaller time interval. All reactions were modelled using this approach, and the model was fitted to the experimental data by adjusting the rate constants to give the minimum sum of squares.
In the special case of reductive amination with chitin or chitosan oligosaccharides prepared by nitrous acid degradation, only Schiff bases (oximes/hydrazones) can be formed. Hence, the general reaction scheme for these reactions is simplified as showed in Figure S6.

S4 Modelling of A n M conjugation reactions
Conjugation reactions (no reducing agent present) with A n M oligomers were studied by time course NMR (S2). The experimental data obtained for the conjugation reactions were fitted using the model described in S3 (based on the reaction scheme presented in Figure S6). Examples of the data fitting for the conjugation of A 2 M oligomers to ADH and PDHA (2 equivalents) at pH 4.0 are given in Figure S7.

S5 Spontaneous decomposition of reducing agents
The decomposition of α-picoline borane (PB) and sodium cyanoborohydride (NaCNBH 3 ) in 500 mM deuterated NaAc-buffer, pH 4.0, was monitored by NMR. The two reducing agents have distinct resonances in the 1 H-NMR spectrum and hence, their decomposition was studied. In contrast to NaCNBH 3 , which is completely dissolved in the buffer, PB has low initial solubility, but dissolves slowly over time (observed by increased intensity of the resonances over time).
The protons of the pyridine ring in PB give resonances with chemical shifts in the range 7.2 to 8.7 ppm, whereas the protons of the methyl group give one resonance with a chemical shift of approximately 2.6 ppm ( Figure S8). The protons of the -BH 3 group are not visible in the 1 H-NMR spectrum due to negative chemical shifts. When PB is oxidised (o), the proton resonances are moved slightly downfield as seen in Figure S8. The relative reductive power (%) for PB over time was calculated by relating the intensity of the protons resulting from the reduced form of PB (r) to the total amount of dissolved PB (r+o) ( Figure S8).
In contrast to PB, the proton resonances of the -BH 3 group in NaCNBH 3 are within the NMR scale (0 -8 ppm). NaCNBH 3 lacks other protons and hence, it was not possible to study the oxidation of this reducing agent. Therefore, the decomposition was related to the reduced intensity of the resonances from the protons in the -BH 3 group relative to an impurity in the buffer ( Figure S9). As the first spectrum was obtained after one hour, the intensity of the resonances at t = 0 was obtained by extrapolation. The relative reductive power (%) for NaCNBH 3 was related to the decomposition of the reducing agent. By comparing the change in relative reductive power (%) for the two reducing agents over time (Figure S10), NaCNBH 3 was shown to decompose approximately 20 times faster than PB in the buffer.

S6 Reduction of A n M oligomers
Reduction of A n M oligomers was performed by adding 3 equivalents reducing agent (PB or NaCNBH 3 ) to oligomers dissolved in deuterated NaAc-buffer. The course of the reduction was studied by monitoring the disappearance of reducing end gem-diol resonances (H1, M and H1, M').
Resonance intensities were related to the internal standard (TSP). Reduction of A n M oligomers by PB was studied at pH 3.0, 4.0 or 5.0 (Figure S11-S13) or by NaCNBH 3 at pH 4.0 ( Figure S14).
For comparison, reduction of AA oligomers (normal reducing end) was studied using 3 equivalents PB (Figure S15) or NaCNBH 3 ( Figure S16) at pH 4.0. In contrast to the A n M oligomers, no detectable reduction in the intensity of the reducing end resonances for AA was observed. However, the relative ratio of the α-to β-reducing end resonances gradually changed ( Figure S15). For this experiment AA was dissolved in the buffer shortly before PB was added and the first spectrum was obtained.

S7 Reduction of A n M conjugates
Reduction of A n M conjugates was studied by adding 3 equivalents reducing agent (PB or NaCNBH  Figure S17 and S18, respectively. The kinetics of the reduction is given in Figure S19 a    Figure S19: Kinetics of the reaction obtained from the spectra in Figure S17 and S18 for the reduction of a) A 2 M-ADH (hydrazone) conjugates and b) A 2 M-PDHA (oxime) conjugates by 3 equivalents PB at pH 4.0, RT, respectively.

S8 Modelling of A n M reduction reactions
Using the rate constants and equilibrium yields obtained for the conjugation of A n M to 2 equivalents ADH or PDHA, the experimental data obtained for the reduction of conjugates (studied by time course NMR, S7) were fitted using the model described in S3 (based on the reaction scheme presented in Figure S6). Examples for the data fitting for the reduction A 2 M-ADH or A 2 M-PDHA conjugates using 3 equivalents PB at pH 4.0 (RT) are given in Figure S20 a

S9 Optimisation of preparative protocols Statistical distribution of mono-, di-and unsubstituted ADH and PDHA
The relative amount of unsubstituted, monosubstituted and disubstituted ADH and PDHA can be calculated assuming the reactivities of the two termini are identical. Such estimates are primarily intended to determine how much ADH or PDHA should be used for mono-substitution (activation) without producing too much disubstituted species.
Definitions and main relations: Let a be defined as the equivalence of linker (ADH or PDHA) relative to oligosaccharide before reaction. Hence: Since we treat each terminal amine as a separate and independent reactant we may further write: Let b be defined as the fraction of oligosaccharide that has become substituted. b is obtained directly from the equilibrium yields listed in Table 1 Hence: The probability that both ends are substituted, fraction of disubstituted species (f DS ) = p 2 The probability that none of the ends are substituted, fraction of unsubstituted species (f US ) = (1 -p) 2 The probability that one end is substituted, fraction of monosubstituted species (f MS ) = 1 -p 2 -(1 - b is obtained directly from the equilibrium yields given in Table 1

Minimisation of disubstituted ADH or PDHA
As shown above the statistical amount of disubstituted ADH or PDHA is significant (5-6 %) when 2 equivalents are used, especially for high reaction yields, but decreases below 1% for 10 equivalents.

Optimisation of reduction conditions for PDHA conjugates
The reduction of PDHA conjugates was optimised by varying the concentration of reducing agent (PB). However, due to the low solubility of PB in the buffer, reduction of A 4 M-PDHA conjugates (equilibrium mixture with 10 equivalents PDHA) using 20 equivalents PB was performed in deuterated buffer in a separate vial (not in the NMR tube). The reaction was performed on a shaking device to increase the collision frequency of undissolved reducing agent and conjugates. NMR spectra of the dissolved phase of the reaction mixture was obtained after 24 and 48 hours revealing complete reduction after 48 hours.

Preparation of A n M conjugates using optimised protocols
Reduced A n M conjugates (A n M-ADH and A n M-PDHA) were prepared using optimised protocols. In brief, conjugation was carried out for 6 hours at RT using 10 equivalents ADH and PDHA at pH 4.0.
For ADH conjugates, reduction was performed for 24 hours at RT by adding 3 equivalents PB to the equilibrium mixture, whereas for the PDHA conjugates, 20 equivalents PB were added to the equilibrium mixture and the reduction was performed for 48 hours at RT to ensure complete reduction. Reaction mixtures were fractionated by GFC and purified conjugates were characterized by 1 Figure S25.  Figure S26 and S27, respectively.

S10 2D NMR characterization of the reduced and purified A 2 M-PDHA
The reduced and purified A 2 M-PDHA conjugate was characterized by homo-and heteronuclear NMR correlation experiments. The conjugate was dissolved in D 2 O and the NMR analysis was carried out using the 800 MHz spectrometer in a 3 mm NMR tube. Resonances were assigned by starting at the anomeric proton signal and then following the proton-proton connectivity using TOCSY, DQF-COSY/IP-COSY, 13 C H2BC and 13 C HSQC-[ 1 H, 1 H] TOCSY spectra. 13 C-HSQC was used for assigning the carbon chemical shifts. The 13 C HMBC spectrum provided information of connections between the sugars. One of the alternative forms of the M residue was structurally elucidated to be 3,5-anhydro-D-mannose. The following designations are used in the spectra displayed in Figure S28 and S29 and Table S1: Table   S1. The structure of the A 2 M-PDHA conjugate is included in Figure S29.

S11 Preparation of chitin-b-chitin diblocks from activated chitin oligomers (A 4 M-ADH)
Chitin-b-chitin diblocks were prepared by reacting A 4 M-ADH conjugates (reduced and purified) with A 4 M oligomers in an equimolar molar ratio. The conjugation of the second block (A 4 M) was monitored by NMR (as described in S2). The kinetics was compared to simulated values for the corresponding reaction using rate constants obtained for the conjugation of A 2 M and A 5 M to free ADH. The results are summarised in Table S2 and kinetic plots are given in Figure S30. The comparison suggests the second conjugation proceeds somewhat faster than the first.  Reduction of the obtained equilibrium mixture was performed using 3 equivalents PB for 24 hours (RT). The reaction mixture was fractionated by GFC ( Figure S31) and main products were purified. oligomers of lower and higher DP), some longer and shorter diblocks were consequently formed. 1 Figure S31.

S12 Preparation of chitin-b-chitin diblocks using a sub-stoichiometric amount of ADH or PDHA
Chitin-b-chitin diblocks were prepared using A 2 M oligomers and a sub-stoichiometric amount of ADH or PDHA (0.5 equivalents). The conjugation of oligomers was monitored by NMR. The kinetics was compared to simulated values using rate constants obtained for the conjugation of A 2 M to free ADH or PDHA (2 equivalents). The results are summarised in Table S3 and the kinetic plots are given in Figure S34.

S13 Preparation and characterization of dextran (Dext m ) oligomers
Dextran (Dext m ) oligomers were prepared by acid hydrolysis of dextran (Dextran T-2000, M w = 2,000,000). 1 H-NMR characterisation of Dextran T-2000 is given in Figure S41. The degree of branching was estimated from the integral of the H1 resonance of internal glucose residues in the main chain and the H1 of the glucose residues at the branching points (BP) to 3.6 % 6 . The degradation mixture was fractionated by preparative GFC to obtain isolated Dext m oligomers ( Figure   S42a). The isolated Dext 6 oligomer (DP = 6) was characterized by 1 H-NMR ( Figure S43) and fractionated by analytical GFC ( Figure S42b) to show the slight polydispersity of the oligomer. The degree of branching of the oligomer was estimated to approximately 0.8 % showing that the branches are more rapidly hydrolysed than the linkages in the main chain.    Figure S42a.

S14 Conjugation of Dext m oligomers to ADH and PDHA studied by time course NMR
Conjugation of Dext 5 oligomers (DP = 5) to ADH and PDHA (2 equivalents) at pH 4.0, RT was monitored by NMR ( Figure S44 and S45, respectively). Here, the reducing end of dextran (Glc, normal reducing end), governs the conjugation and hence, the acyclic hydrazones and oximes are in equilibrium with cyclic N-glycosides. As previously shown 5 , dextran formed almost exclusively Npyranosides with ADH, whereas it formed N-pyranosides in addition to E-and Z-oximes with PDHA.
The equilibrium yield of conjugates (E-/Z-hydrazones or oximes + N-pyranosides) obtained for the reactions was 35% for Dext 5 with ADH and 87% for Dext 5 with PDHA. The experimental data obtained in the conjugation reactions were fitted using the model described in S3 (based on the reaction scheme presented in Figure S5). The data fitting for the conjugation Dext 5 oligomers to ADH and PDHA (2 equivalents) at pH 4.0 (RT) are given in Figure S46. Rates (t 0.5 and t 0.9 ) and equilibrium yields for the total conjugation reaction are given in Table S4.

S15 Preparation of chitin-b-dextran diblocks
Chitin-b-dextran diblocks were prepared by reacting A 5 M-ADH or A 5 M-PDHA conjugates (reduced and purified) with dextran oligomers of DP = 6 (Dext 6 ) in an equimolar ratio. The conjugation of the dextran block was monitored by NMR (as described in S2) and combined equilibrium yields of 15 and 66 % were obtained for the conjugation to A 5 M-ADH or A 5 M-PDHA, respectively. The kinetics was (as above) compared to simulated values using rate constants (k 1 , k -1 etc) obtained for the conjugation of Dext 5 to free ADH or PDHA. The results are summarised in Table S5 and the kinetic plots are given in Figure S47. Compared to the model, it appears the second conjugation is faster and gives higher yields than the first.  diblocks, respectively, due to the slower reduction of ADH conjugates 5 . The reaction mixtures were fractionated by GFC ( Figure S48), and main fractions were purified and characterized by 1 H-NMR ( Figure S49-S53). As purified A 5 M conjugates were used for the diblock preparation, the integral for the H1, A resonance was set to 5 in all the 1 H-NMR spectra. The yield of diblocks was obtained by integrating the chromatograms ( Figure S48) as described above. Due to the slight polydispersity of the Dext 6 oligomer, some longer and shorter diblocks were also formed. The weight yield of diblocks was 85 and 92 % for chitin-b-dextran diblocks with ADH and PDHA, respectively. The amount of remaining unreduced diblock could not be accurately determined because the resonance corresponding to the unreduced N-pyranoside conjugates overlaps with other resonances in the spectra. However, the integral for the secondary amine resonance balanced the integrals of the resonances from both chitin and dextran, suggesting close to complete reduction of diblocks.
Approximately 40 % the unreacted dextran oligomers from the reaction with A 5 M-ADH conjugates (20 equivalents PB, 40 °C) were reduced after 144 hours (6 days) ( Figure S51). Hence, the dextran oligomers are reduced by PB with a much slower rate than the A n M oligomers. The slow reduction of dextran oligomers also explains the higher yield of diblocks obtained compared to the low amination yield, as the dextran oligomers can react further after addition of reducing agent.       Figure S48b.