*2.6. Specific Mode of Action of ScCDA2 on A4*

The partially acetylated chitooligosaccharide derivatized with a reducing amine showed molecular weights of 1005.96 Da (A3D1), 963.82 Da (A2D2) and 921.80 Da (A1D3) in MALDI-TOF mass spectrometry (Figure 7).

**Figure 7.** Analysis of partially acetylated COS by reductive amine derivatization with mass spectrometry. The reaction product generated after *Sc*CDA2 treatment (GlcNAc)4 was labelled with AMAC and analyzed by MALDI-TOF-MS. (**A**). MS1 spectrum of A3D1 labelled with AMAC (*m*/*z* 1005.96). (**B**). MS1 spectrum of A2D2 labelled with AMAC *m*/*z* 963.82). (**C**). MS1 spectrum of A1D3 labelled with AMAC (*m*/*z* 921.80).

Then, by applying MALDI-TOF-MS analysis in MS2 mode, we were able to identify and analyze *Sc*CDA2's partially acetylated products and determine specific acetylation pattern of partially acetylated chitooligosaccharides. As is shown in Figure 8, the A4 is first deacetylated to DAAA, ADAA and AADA (Figure 8A), followed by further deacetylation products to the intermediate product DDAA, DADA and ADDA (Figure 8B). Finally, the end product DDDA was obtained, due to the

third deacetylation (Figure 8C). Therefore, deacetylation occurred mainly at the non-reducing end, and the acetyl at the reducing end was always present. No matter how we prolonged the reaction time or increased the concentration of the enzyme, the acetyl group at the reducing end could not be removed. After comparing the intermediate and end products generated by the deacetylation of the chitin tetramer, we found that the deacetylation occurred at any position except for the reducing end, indicating that *Sc*CDA2 has a multiple attack mechanism like *Cl*CDA and *Sp*PgdA [33,36]. However, *Sc*CDA2 cannot deacetylate at the reducing end to form completely deacetylated COS (DDDD). Therefore, the deacetylation pattern of *Sc*CDA2 is significantly different from the reported CDA derived from *C. lindemuthianum*, *Mucor rouxii*, *Aspergillus nidulans*, *Vibrio cholera*, *Puccinia graminis*, etc [14,29,30,41,47]. A "subsite-capping model" has been proposed to explain the differentiation of the deacetylation process and product patterns of CDA [30]. This subsite-capping model states that the position and dynamics of loops play an important role in substrate preference and regioselectivity of deacetylation. So, the difference in the deacetylation mode of *Sc*CDA2 may be due to the loop length, position and dynamic effects [47].

**Figure 8.** MALDI-TOF-MS2 determines the acetylation pattern of partially acetylated COS. (**A**) The MS2 spectrum of A3D1 labelled with AMAC and the resulting ion fragments: A-amac, DA-amac, AA-amac, ADA-amac, DAA-amac, AAA-amac (*m*/*z* 438.15, 599.19, 641.20, 802.23, 844.24); so, the acetylation pattern of A3D1 is DAAA, ADAA and AADA. (**B**) MS2 spectrum of A2D2 labelled with AMAC and the resulting ion fragments: A-amac, DDA, DAD, ADD, DA-amac, AA-amac and ADA-amac, (*m*/*z* 438.16, 548.19, 599.22, 641.22, 802.29); so, the acetylation pattern of A2D2 is DDAA, ADDA and DADA. (**C**) MS2 spectrum of A1D3 labelled with AMAC, resulting in ion fragmentation of A-amac, DA-amac and DDA-amac (*m*/*z* 438.16, 599.21, 760.25); so, the acetylation pattern of A1D3 is DDDA. (**D**) The deacetylation process of *Sc*CDA2 when A4 is used as a substrate.
