Mass spectrometry imaging of oligosaccharides following in situ enzymatic treatment of maize kernels

Carbohydrate Polymers 275 (2022) 118693

Contents lists available at ScienceDirect

Carbohydrate Polymers
journal homepage: www.elsevier.com/locate/carbpol

Mass spectrometry imaging of oligosaccharides following in situ enzymatic
treatment of maize kernels
Jonatan R. Granborg a, b, *, Svend G. Kaasgaard b, Christian Janfelt a
a
b

Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark
Novozymes A/S, Biologiens Vej 2, 2800 Kongens Lyngby, Denmark

A R T I C L E I N F O

A B S T R A C T

Keywords:
Polysaccharides
Mass spectrometry imaging
Oligosaccharides
Enzymes
Degradation
Maize

In recent years enzymatic treatment of maize has been utilized in the wet-milling process to increase the yield of
extracted starch, proteins, and other constituents. One of the strategies to obtain this goal is to add enzymes that
break down insoluble cell-wall polysaccharides which would otherwise entrap starch granules. Due to the high

complexity of maize polysaccharides, this goal is not easily achieved and more knowledge about the substrate
and enzyme performances is needed. To gather information of both enzyme performance and increase substrate
understanding, a method was developed using mass spectrometry imaging (MSI) to analyze degradation products
from polysaccharides following enzymatic treatment of the maize endosperm. Different enzymes were spotted
onto cryosections of maize kernels which had been pre-treated with an amylase to remove starch. The cry­
osections were then incubated for 17 h. before mass spectrometry images were generated with a MALDI-MSI
setup. The images showed varying degradation products for the different enzymes observed as pentose oligo­
saccharides differing with regards to sidechains and the number of linked pentoses. The method proved suitable
for identifying the reaction products formed after reaction with different xylanases and arabinofuranosidases and
for characterization of the complex arabinoxylan substrate in the maize kernel.
Hypotheses: Mass spectrometry imaging can be a useful analytical tool for obtaining information of poly­
saccharide constituents and enzyme performance from maize samples.

1. Introduction
Xylans are one of the most widespread polysaccharides found in
terrestrial plants differing in structure and complexity, from linear
homoxylans to complex arabinoxylans decorated with a variety of
sidechains (Bajpai, 1997). Due to its abundancy in the plant kingdom,
xylans and other hemicelluloses have an impact on many everyday
products and industrial processes, knowledge of their structure, location
and function in plants is therefore an important research field (Biely,
´, Hroma
´dkova
´, & Heinze, 2005).
Singh, & Puchart, 2016; Ebringerova
In the wet-milling process, starch and other valuable products, such
as protein, oil and fiber are isolated from the maize kernels (Rausch,
Hummel, Johnson, & May, 2019). The wet-milling industry is very large

with approximately 40.6 Mt. maize being processed in 2015 (Rose­

ntrater & Evers, 2018). In cereal grains, hemicelluloses have been shown
to differ significantly with regards to structure and function depending
´, 2005) and maize have
of the type of grain(Bajpai, 1997; Ebringerova
some of the most complex cell-wall polysaccharides with highly deco­
rated arabinoxylans(Bajpai, 1997; Rosicka-Kaczmarek, Komisarczyk,
Nebesny, & Makowski, 2016). A strategy to allow an efficient extraction
of the starch and protein from the kernels, is to open or remove the
structures of the cell-wall polysaccharides, protein-related structures,
and other polymers in the endosperm (Ozturk, Kaasgaard, Palm´en,
Vidal, & Hamaker, 2021; P´
erez-Carrillo & Serna-Saldívar, 2006).
A method for increasing the yield of starch, and possible other
components, obtained from the classical mechanical wet-milling process

Abbreviations: 8–5′ -DFAdc, Decarboxylated 8–5′ -dehydroferulic acid; Ac, Acetyl group; Araf, Arabinofuranosyl residue; CE, Carbohydrate esterase; diFA, dimer of
ferulic acid; FA, Ferulic acid; Galp, Galactopyranosyl; GH, Glycoside hydrolase; Glcp, Glucopyranosyl residue; GlcA, Glucuronic acid; Hex, Hexose; HexA, Hexuronic
acid; MeGlcA, 4-O-Methyl-glucuronic acid; MeHexA, 4-O-methyl-hexuronic acid; MeHexAn, Oligosaccharide composed of n numbers of methyl-hexuronic acids; Pen,
Pentose; Penn, Oligosaccharide composed of n numbers of pentoses; PBOS, Pentose-based oligosaccharides; Xylp, Xylopyranosyl residue.
* Corresponding author at: Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen,
Denmark
E-mail addresses: , (J.R. Granborg).
/>Received 9 July 2021; Received in revised form 15 September 2021; Accepted 19 September 2021
Available online 23 September 2021
0144-8617/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license ( />

J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

Fig. 1. Scheme of a possible corn arabinoxylan inspired by Agger et al. and Biely et al. (Agger, Viksø-Nielsen, & Meyer, 2010; Biely et al., 2016).

is through enzymatic degradation of cell-wall polysaccharide leading to
an increased release of starch from the endosperm of the maize kernel
(P´erez-Carrillo & Serna-Saldívar, 2006). The most abundant of nonstarch polysaccharides in maize are arabinoxylans composed of back­
bone of β-d-xylopyranose linked by β (1 → 4) glycosidic bonds decorated
with arabinose units binding on C2 and/or C3 of the xylose residue.
Apart from arabinose units, the xylan backbone of maize arabinoxylan
can also be decorated by other compounds including; ferulic acid (FA),
also as cross-linking dimers and trimers (Bento-Silva, Vaz Patto, &
Ros´
ario Bronze, 2018), binding on C5 of the arabinose units, hexuronic
acids (HexA) and 4-O-methylhexuronic acids (MeHexA) binding on C2
and acetyl groups, binding on O-2 and/or O-3 (Chesson, Gordon, &
Lomax, 1983; Coelho, Rocha, Moreira, Domingues, & Coimbra, 2016;
Huisman, Schols, & Voragen, 2000; Saulnier, Marot, Chanliaud, &
Thibault, 1995). The exact structure of arabinoxylans from maize

endosperm has not yet been determined, but is theorized to be highly
branched by arabinose units attached to the xylose backbone and in
addition may contain both ferulic acid (FA) and HexA units (ChateignerBoutin et al., 2016), similarly to the xylan species of the pericarp. In
Fig. 1 a model structure of maize arabinoxylan is depicted, demon­
strating some of the possible branching units based on arabinoxylan
structures found in the corn pericarp and fiber. Due to the complex
nature of the maize arabinoxylans, more knowledge is needed about the
possible structural components of the arabinoxylans to find more spe­
cific enzymes or combinations of enzymes to enhance enzymatic
degradation.
In this study, a method based on mass spectrometry imaging was
developed for measuring the degradation products of enzymes with

different specificities, either alone or in combinations, from the endo­
sperm of maize kernels. Mass spectrometry imaging (MSI) is an

Fig. 2. Sketch of sample preparation workflow for MSI analysis of enzyme treated maize kernels. Step 1: Rehydration overnight of dried maize kernel. Step 2:
Cryosectioning of maize kernel embedded in 10% gelatin. Step 3: Enzymatic starch removal with α-amylase (GH13_1). Step 4: Incubation with different enzymes at
fixed humidity. Step 5: MALDI imaging experiment. Step 6: Data analysis and image generation.
2


J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

analytical technique which couples the chemical selectivity of mass
spectrometry with spatial information obtained by measuring in a grid
across the surface of the sample. While MSI can be used for targeted
analysis of selected compounds, one of the greatest strengths of the
technique is its ability to track the distribution of multiple analytes
simultaneously across a sample. This ability makes MSI ideal for the
analysis of enzymatic degradation products from complex heterogenous
substrates such as the carbohydrates analyzed in this study.
While cell-wall polysaccharides and enzymatic treatment in cereal
grains have been described and shown with MSI before (Fanuel,
Ropartz, Guillon, Saulnier, & Rogniaux, 2018; Feenstra et al., 2017;
ălling, Nattkemper, & Niehaus, 2016; Peukert, Lim, Seiffert,
Gorzolka, Ko
& Matros, 2016; Veliˇckovi´c et al., 2016), focus on comparing enzymes
by the specific oligosaccharide degradation products they produce after
in situ application is new. Additionally, specific oligosaccharide knowl­
edge obtained with this technique functions as building blocks that,

when put together, can be used to describe the substrate in more detail.

protect it from droplets formed due to condensation on the lid of the
container. After incubation, the samples were dried for 30 min in a
vacuum desiccator.
2.5. MALDI-MSI
A matrix solution of 20 mg/mL THAP in 90:10 methanol:H2O (v/v)
was applied using an iMatrixspray (Stoeckli, Staab, Wetzel, & Brech­
buehl, 2014). The parameters were set to spray 14 cycles with a line
distance of 1 mm on an area of 40 × 40 mm2 from a height of 80 mm at a
speed of 90 mm/s with a density of 4 μL/cm2 and with no delay between
cycles.
The MSI analysis was performed at ambient conditions with a
SMALDI10 ion source (TransMIT, Giessen, Germany) attached to a
Thermo Q-Exactive Orbitrap mass spectrometer (Thermo Scientific,
Bremen, Germany) operated at a mass resolving power of 140,000 at m/
z 200. The analysis was done in positive ion mode with a scan range of
300–2100 m/z and a scan speed of around 1 pixel/s. The pixel size was
set to 30 μm resulting in an average image size of approx. 65,000 pixels,
step 5 in Fig. 2.

2. Experimental
2.1. Chemicals

2.6. MALDI image generation

The matrix used was 2′ ,4′ ,6′ -Trihydroxyacetophenone monohydrate
(THAP) acquired from Sigma-Aldrich. For embedding a food-grade
gelatin powder was used (Haugen-Gruppen Nordic A/S). Water was
prepared with a Millipore Direct-Q3 UV system (Billerica, MA, USA).

The mass spectra obtained from the MALDI experiment were coupled
with their spatial location by conversion to imzML files (Schramm et al.,
2012) and analyzed in MSiReader v1.02 (Bokhart, Nazari, Garrard, &
Muddiman, 2018; Robichaud, Garrard, Barry, & Muddiman, 2013). The
m/z-tolerance was set to ±4 ppm and the only pre-processing of the
samples was normalization to Total Ion Current (TIC), step 6 in Fig. 2.

2.2. Enzymes
All enzymes were supplied by Novozymes A/S (Lyngby, Denmark)
and they were all purified to homogeneity.

2.7. Data analysis

2.3. Maize sample preparation

Due to the number of possible oligosaccharides from the maize
hemicelluloses, a targeted analysis approach was used to assess the MSI
data. Targeted analysis in this case means that the compounds searched
for were all based on reports of compounds found in the literature,
assumed degradation products based on the proposed structure of
polysaccharides, in-house knowledge, and combinations of compounds.
For this purpose, a library of possible oligosaccharides that might be
available with and without enzymatic treatment of the maize kernel was
compiled by calculating accurate masses of >7000 compounds within in
the m/z-range (300− 2100). The compounds were generally structured
around oligosaccharides based on a main component, primarily pentose,
with one or more sidechain compounds attached, such as ferulic acid,
acetyl groups, dimers of ferulic acid, hexuronic acid and methylated
hexuronic acid. For all the compounds the m/z values for the following

ions: [M + H]+, [M + Na]+, [M + K]+, [M + H - H2O]+, [M + Na - H2O]+
and [M + K - H2O]+ were included in the library. In untreated samples,
the dominant adduct ion was found to be the potassium adduct. After
enzymatic treatment, the sodium adduct without a neutral loss of water
was found to be the dominant ion adduct. The shift from the potassium
adduct to the sodium adduct is believed to be caused by excess amount
of sodium in the enzyme solutions received as no sodium was added in
the following sample preparation.
Due to the complexity of the oligosaccharides from cell-wall poly­
saccharides and yet structural similarity, there were several isomeric
compounds e.g., Pen4 has exactly the same m/z-value as Hex3-Ac.
Because of such isomers, the number of unique m/z-values amounted to
only >3000. However, because of the mass resolving power of the QExactive mass spectrometer it was possible to differentiate between most
other compounds. More specifically it was possible to differentiate be­
tween oligosaccharides having up to 12 pentose units with a HexA or FA
sidechain although they differ only by approximately 0.015 in m/z
value. As an additional precaution to minimize the risk of misinterpre­
tation of the data, the compounds were generally looked for as a series of
oligosaccharides and only used when a sufficient pattern was observed.

Flint corn kernels provided by Novozymes North America Inc.
(Franklinton, NC, USA) were used for the experiments. Before further
preparation of the sample, the kernels were rehydrated in Milli-Q water
for 17 h. at ambient conditions, step 1 in Fig. 2.
The rehydrated maize kernels were embedded in a 10% w/v gelatin
gel using a rubber mold. After embedding, the mold containing the
sample was transferred to − 80 ◦ C until cryosectioning. Cryosectioning
was performed on a Leica CM3050S cryo-microtome (Leica Micro­
systems, Wetzlar, Germany) at − 23 ◦ C using Kawamoto Cryotape 2C(9)
(SECTION-LAB Co. Ltd. Hiroshima, Japan) to achieve 10 μm sections of

the fragile sample (Kawamoto & Kawamoto, 2021). The sections were
then attached to a standard microscope slide using a double-sided ad­
hesive carbon tape (SPI supplies) and stored at − 80 ◦ C until use, step 2 in
Fig. 2.
The samples were transferred from the − 80 ◦ C freezer to a vacuum
desiccator to dry the sample to limit dislocation of analytes. After dry­
ing, the samples were incubated 24 h. at 40 ◦ C in a 50 mL falcon tube
containing a 100 ppm solution of α-amylase (GH13_1) in 10 mM
ammonium acetate buffer at pH ~ 5, step 3 in Fig. 2. After incubation,
the slide was rinsed with Milli-Q water twice before being left in a fume
hood to dry for 2 h. This step was introduced to reduce ion suppression
by removal of starch.
2.4. Enzyme treatment
When the samples had dried, three droplets of 2 μL 100 ppm enzyme
in 10 mM ammonium acetate buffer were applied to the endosperm and
the slide was transferred to a plateau in a plastic box which was sealed
and incubated for 17 h. at 40 ◦ C. To ensure that the enzymes would not
dry out during the incubation, 75 mL of saturated K2SO4 solution was
added to the box keeping the relative humidity in the headspace at 96%,
step 4 in Fig. 2. A microscope slide was positioned above the sample to
3


J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

Due to the focus of looking for numerous possible degradation products
after enzymatic treatment it was not possible to also obtain MS/MS data
and thereby further elucidate the chemical structures of the individual

oligosaccharides.
To create an overview of enzyme specific patterns for degradation
products, abundance data as TIC for >1500 oligosaccharides were
exported for regions of interest (ROI) corresponding to the spatial
location of each enzyme, example of a ROI is shown in Fig. S1. The
abundance data were then organized in JMP®, Version 14.1. SAS
Institute Inc., Cary, NC, 1989–2019. To track the location of the droplet
deposited enzyme in the endosperm of the maize, MS images were
generated of the buffers that the enzymes were dissolved in when
received. The buffers, HEPES-Na (as trimer) and MES-Na (as dimer),
effectively showed the area covered by the droplets, see Fig. S2. To in­
crease the visual comparison when studying multiple MS images, the
imzML-files were merged together and fitted to one intensity scale.

the pentose monomers of each oligosaccharide. In Fig. 3, the cumulative
abundance of PBOS made of between two and ten pentoses in the
endosperm of maize kernels following enzyme treatment are shown. It is
evident from Fig. 3, that the availability of specific oligosaccharides was
detectable and assignable to areas treated with specific enzymes. It was
also possible to differentiate between the enzyme solutions applied in
terms of immediate effectiveness with regards to production of the
studied degradation products.
For the PBOS consisting only of pentoses seen in Fig. 3, it is evident
that the GH5 type enzymes seen in section 3 and 4 to a higher degree
produced these oligosaccharides than GH10, even when combined with
arabinofuranosidase as shown in section 1 and 2. This effect was not
limited to the pentose-only oligosaccharides, similar effects were
observed for more decorated oligosaccharides. Some of these are shown
in Fig. 4, where the distribution of PBOS with the most abundantly
observed decorations are presented for endosperm treated with either

GH10 + GH62 (section 2) or GH5_21 (section 3). The findings of various
oligosaccharides decorated by one or more sidechains of FA, HexA,
MeHexA and especially Ac-groups corresponds well to other findings
within the field (Appeldoorn, Kabel, Van Eylen, Gruppen, & Schols,
2010). It was however interesting that the most predominantly ionized
type of oligosaccharides based on Fig. 4.B were PBOS having both an
acetyl group and a hexuronic acid residue attached.
The abundancy of PBOS with additional sidechains showed the ne­
cessity for a broad analysis when determining degradation products
from complex carbohydrates such as cell-wall polysaccharides in maize
endosperm. A noteworthy limitation to this widespread analysis when
comparing enzyme performance is that the measurements are at best
semi-quantitative and can only give an indication of the relative
amounts of oligosaccharides. To reduce ion suppression, it was neces­
sary to remove starch before incubating samples with the various xyla­
nases. During this amylase treatment and following wash steps, most of
the soluble arabinoxylans is also removed. However, as the focus of the
study was to evaluate the ability of different enzymes to open poly­
saccharides structures this loss of inherently available oligosaccharides
was considered beneficial as it reduced the background oligosaccharide
signal.

3. Results and discussion
3.1. Degradation products after in situ enzyme treatment
Due to the complex hemicellulotic structures in the maize kernels, a
variety of oligosaccharides can be formed as degradation products
following enzymatic treatment. A thorough analysis of thousands of m/
z-values corresponding to oligosaccharides of several chemical species
with or without sidechain decorations was performed in this study. Since
the distributions of all the compounds are extractable from each indi­

vidual MS image the results presented are all obtained from the four
maize sections shown in Fig. 3.
The simplest pentose-based oligosaccharides (PBOS) are made
entirely of pentoses with no other compounds attached. The pentose
monomers can, depending on the type of polysaccharide, consist of
various pentoses, for maize arabinoxylans these are composed of a xylan
backbone where some of the xylose residues can be linked to an arabi­
nose residue in O2 and/or O3 position. Since the focus of this study was
to identify different oligosaccharides from each enzyme experiment
rather than distinguishing between the type of and structural location of

Fig. 3. Left: Combined image of four MALDI imaging experiments of maize kernels, each treated with three different enzyme solutions, showing the cumulative
intensity for Pen2-Pen10 as a percentage of TIC. The image show that oligosaccharides made of pentoses without sidechains are predominantly found when the
endosperm is treated with the enzymes GH5_21 or GH5_34. Individual MS images for Pen2-Pen10 can be seen in the supporting information Fig. S3. Right: Photo of
the third section after 2 μL of enzyme solution was applied.

4


J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

Fig. 4. Top: Chart depicting the mean intensities of PBOS with different sidechains for a ROI corresponding to an area of an endosperm in a section of a maize kernel
treated with GH10 + GH62. Bottom: Chart depicting the mean intensities of PBOS with different sidechains for a ROI corresponding to an area of a treated endosperm
in section of a maize kernel treated with GH5_21. The images in the chart uses an MS image from Fig. S1 to show the area of the section treated with the
enzyme solution.

3.2. GH10 alone and combined with other enzymes

results for the PBOS observed in this study with GH10 was in combi­
nation with GH62. Interestingly, the GH62 arabinofuranosidase showed
significantly better results than the GH43 arabinofuranosidase, when
combined with GH10, suggesting that xylose residues with arabinose
sidechain are mainly present in monosubstituted form. Furthermore, the
results in Table 2 show that while the method is only to be considered as
semi-quantitative, the profiles obtained are reproducible with similar
patterns observed for both GH10 + GH62 droplets.
The blend of GH10 and GH62 showed an interesting synergistic ef­
fect. This can be seen in Fig. 5, in which both the chart and the MALDI
image clearly show a production of ferulic acid PBOS by the combina­
tion of GH10 and GH62 whereas these oligosaccharides were not formed
when the enzymes were added separately. This supports the finding of
Puchart et al. (Puchart et al., 2007), that GH62s cannot hydrolyze
substituted Araf units including feruloyated and acetylated decorations.
While synergetic effects have been found before for Endo-1,4-β-xyla­
nases in combination with α-L-arabinofuranosidases (Ravn et al., 2018),
the specific distribution of the individual oligosaccharides produced has

The endo-1,4-β-xylanase GH10, is a xylanase which is dependent on
two connected unsubstituted xylose units to cleave a xylan backbone
(Biely et al., 2016). In this study GH10 was used to evaluate if the cellwall polysaccharides from the endosperm of maize kernels consistently
were too complex for this class of xylanases. It was also used in com­
bination with other enzymes to see if removal of specific sidechains
could increase the potential of GH10 for these complex substrates. In
section 1 (see Fig. 3) GH10 and GH62 were applied separately or in
combination to find out if removal of arabinose residues from the
backbone of polysaccharides would increase the activity of GH10. In
section 2 (see Fig. 3), the combination of GH10 and GH62 was compared
to combinations of GH10 with another α-L-arabinofuranosidase, GH43,

and GH10 with an arabinogalactan endo-β-1,4-galactanase, GH53. In
Table 2, a comparison of some of the most prevalent PBOS are shown for
the enzyme combinations from section 1 and 2 (section 3 and 4 are
available in Table S1). This comparison clearly shows that the best

5


J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

Fig. 5. Combined MALDI images from section 1 (i1) and 2 (i2), see Fig. 3, showing the cumulative intensity signal, as a percentage of TIC, for Pen2-FA to Pen8-FA [M
+ Na]+. Chart depicting the distributions of Pen2-FA to Pen8-FA [M + Na]+ based on mean TIC of ROI calculated from more than 1500 pixels. GH10 (Blue), GH10 +
GH62 section 1 (Red), GH62 (Green), GH10 + 62 section 2 (Purple), GH10 + GH53(Brown), GH10 + GH43(Cyan).

not been shown. This information allows for a more specific visualiza­
tion of the building blocks for maize cell-wall polysaccharides and
enzyme specificity. An example of this is seen in Fig. 5, where PBOS with
FA are observed after treatment with GH62. Since ferulic acid is believed
to be exclusively bound to Araf units in arabinoxylan (Malunga & Beta,
2016), the Araf units are not accessible to GH62, otherwise the PBOS
would not have FA attached.

backbones (see Table 1). The dominating PBOS are the pentose based
series with both HexA and an acetyl group attached. Due to the many
PBOS observed with Ac groups in section 3 (Fig. 3), the acetylxylan
esterase, CE5, was applied in combination with GH5_21, and GH5_34, to
see if that would change the degradation product profiles. It was thus
expected that more PBOS without Ac sidechains were formed and the

level of PBOS decorated with Ac groups decreased. Surprisingly, this
effect was not observed in the Penn-series (Fig. 3/S3) where the endo­
sperm treated with GH5_21 and GH5_34 appeared more efficient in the
absence of CE5. For the GH5_34 treated areas, this was also the case for
the Penn-HexA-Ac series (Fig. 6), where GH5_34 alone proved superior
for formation of all other PBOS than Pen3-HexA-Ac and especially for
Pen6-HexA-Ac. For GH5_21, addition of CE5 (Fig. 6, brown) seemed to
create a shift in affinity towards producing PBOS with a shorter chain

3.3. GH5 types with/without CE5
In contrast to GH10, the GH5 types (section 3 and 4 in Figs. 3 and 6)
both appear to produce more and several different types of PBOS, the
most abundantly available are shown for GH5_21 in Fig. 4 as would also
be expected as these xylanases are able to act on highly decorated xylan

6


Carbohydrate Polymers 275 (2022) 118693

J.R. Granborg et al.

Fig. 6. Three MALDI images from enzyme experiments on section 3 (i3) and 4 (i4) with GH30, GH5_34, GH5_21, CE5, CE5 + GH5_21, CE5 + GH5_34 showing the
distribution of the oligosaccharides Pen3-HexA-Ac [M + Na]+ observed at m/z 655.1692 (left), Pen6-HexA-Ac [M + Na]+ observed at m/z 1051.2960 (middle) and
Pen10-HexA-Ac [M + Na]+ observed at m/z 1579.4650 (right) as a percentage of TIC. Chart depicting the distributions of Penn-Hexuronic acid-Ac based on mean TIC
of ROI with more than 1500 pixels from enzyme areas for GH30 (Green), GH5_34 (Red), GH5_21(Blue), CE5 (Purple), CE5 + GH5_21(Brown), CE5 + GH5_34(Cyan).

length for the Penn-HexA-Ac series compared to GH5_21 alone (Fig. 6,
Blue). This effect was particularly notable for the Pen3-HexA-Ac which
was formed in significantly higher levels from the endosperm when

GH5_21 and CE5 was used in combination than when either was used

alone. Similarly, other PBOS were checked to see how CE5 affect the
GH5 type enzymes. For Penn-HexA-Ac-Ac, the effects were largely
negligible (Fig. S4). For Penn-HexA, an increase in signal intensity was
observed, but it accounted for a small part of the total signal and was

Table 1
Enzyme overview.
Enzyme

EC

Organism

Family

Comment

α-amylase

Acetylxylan esterase
Endo-1,4-β-xylanase

3.2.1.1
3.2.1.72
3.2.1.8

GH13_1
CE5

GH10

α-L-arabinofuranosidase

3.2.1.55

GH62

Degrades α-1,4 linkage between adjacent glucose units (Roth et al., 2019).
Degrades mono- and diacetylated xylp residues (Biely et al., 2016).
Cleaves xylan main chain when recognizing two consecutive unsubstituted xylp residues (
Biely et al., 2016).
Removes monosubstituted arabinofuranose (Biely et al., 2016).

Arabinogalactan endo-β-1,4galactanase
α-L-arabinofuranosidase

3.2.1.89

Rhizomucor pusillus
Trichoderma reesei
Talaromyces
leycettanus
Talaromyces
pinophilus
Humicola insolens

GH53

Hydrolyzes β-1,4 galactosidic bonds in arabinogalactan (de Lima et al., 2016).

3.2.1.55

Humicola insolens

GH43_36

Endo-1,4-β-xylanase

3.2.1.8

Bacillus sp

GH30_8

Endo-1,4-β-xylanase

3.2.1.8

Chryseobacterium sp

GH5_21

Arabinoxylanase

3.2.1.-

Gonapodya prolifera

GH5_34

Removes α-1,3 Araf from disubstituted Xylp residues in the xylan backbone (Biely et al.,
2016).
Cleaves glucuronoxylan at the second glycosidic linkage following MeGlcA substituents (Biely
et al., 2016).
Subfamily which has shown xylanolytic effect on wheat arabinoxylan (Dodd, Moon,
Swaminathan, Mackie, & Cann, 2010).
Hydrolyzes highly decorated xylan backbones (Labourel et al., 2016).

7


J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

similar to the effect of CE5 alone (Fig. S5). For Penn-Ac, a decrease in
signal intensity was observed when CE5 was added (Fig. S6), but sur­
prisingly, a corresponding increase of Penn was not observed (Fig. 3/S3).
A possible reason for this inconsistency in the relationship of increase
and decrease in Penn and Penn-Ac, respectively, could be that the
addition of CE5 by debranching the polysaccharides decreases the GH5
type enzymes affinity for some of their cleavage sites. This hypothesis
needs to be confirmed, for example by applying the enzymes sequen­
tially instead of simultaneously.
Another interesting finding for the GH5 type enzymes, with and
without addition of CE5, was the detection of PBOS with a m/z-value
corresponding to the ferulic acid dimer, decarboxylated 8–5′ -dehy­
droferulic acid (8–5′ -DFAdc), previously described by Bento-Silva et al.
(Bento-Silva et al., 2018). Since dimers of FA are expected to crosslink

the arabinoxylan with other arabinoxylans and/or other hemicelluloses
(de O. Buanafina, 2009; Grabber, Ralph, & Hatfield, 2000), this activity
could be of importance for opening the maize cell wall polysaccharide
structures. MALDI images for the four most dominating of these PBOS
can be seen in Fig. 7. A synergetic effect for the combination of GH5_21
and CE5 similar to the one shown for Pen3-HexA-Ac in Fig. 6 was
observed for Pen4–8-5′ -DFAdc.

Table 2
Comparison of enzymatic production of PBOS with various sidechains.
Sidechain

None
Hex
2xHex
Ac
2xAc
MeHexA
FA
2xFA*
HexA
2xHexA
HexAAc**
FA-Ac**
AcMeHexA

Enzymes used for image 1 in
Fig. 3

Enzymes used for image 2 in Fig. 3

GH10

GH62

GH10 +
62

GH10 +
62

GH10 +
43

GH10 +
53

+
+
+
+
+
+
+
+
+
+

+
+

+
+
+
+
+
+
+
+

+
+
+
+++
+
+
++
+
+
+

++
+
+
+++
+
+
++
+
+
+

+
+
+
+
+
+
+
+
+
+

+
+
+
+
+
+
+
++
+
+

++

+

+++

+++

++

++

+

+

+

+

+

+

+

+

+

+

+

+

+: mean abundance<0.01% of TIC.

++: mean abundance>0.01% of TIC.
+++: mean abundance>0.1% of TIC.
*: Pen11 deducted due to overlap with buffer m/z-value.
**: Pen12 deducted due to overlap with buffer m/z-value.

Fig. 7. MALDI images from 2 enzyme experiments with GH30, GH5_34, GH5_21, CE5, CE5 + GH5_21, CE5 + GH5_34. Image 1 shows the distribution of Pen4–8-5′ DFAdc [M + Na]+ observed at m/z 893.2686. Image 2 shows the distribution of Pen5–8-5′ -DFAdc [M + Na]+ observed at m/z 1025.3109. Image 3 shows the dis­
tribution of Pen6–8-5′ -DFAdc [M + Na]+ observed at m/z 1157.3531. Image 4 shows the distribution of Pen7–8-5′ -DFAdc [M + Na]+ observed at m/z 1289.3954.
Intensity shown as a percentage of TIC.
8


J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

Fig. 8. MS Images and chart of section 3 (i3) and 4 (i4) showing the distribution of MeHexAn-Pen1 oligosaccharides in maize sections treated with GH30 (Green),
GH5_34 (Red), GH5_21(Blue), CE5 (Purple), CE5 + GH5_21(Brown), CE5 + GH5_34(Cyan). Image 1: shows the distribution of MeHexA3-Pen1 [M + Na]+ observed at
m/z 743.1853. Image 2: shows the distribution of MeHexA3-Pen1 [M + Na]+ observed at m/z 933.2330. Image 3: shows the distribution of MeHexA5-Pen1 [M + Na]+
observed at m/z 1123.2807. Image 4: shows the distribution of MeHexA6-Pen1 [M + Na]+ observed at m/z 1313.3285. Intensity shown as a percentage of TIC.

3.4. GH30

which can be seen when comparing Penn (Fig. 3/S3) and Penn-Ac (Fig.
S6) with Penn-HexA-Ac (Fig. 6) and Penn-Ac2-HexA (Fig. S4). In addition
to the ability of GH30 to produce PBOS, MeHexAn-based oligosaccha­
rides, both as unsubstituted MeHexAn (Fig. S7) and as MeHexAn-Pen
(Fig. 8), were also unexpectedly observed in maize endosperm after
GH30 treatment. To understand the origin of the MeHexAn-based oli­
gosaccharides further investigations are required. A possibility would be
that GH30 also have a catalytic effect on pectin or pectin-like poly­

saccharides as previously theorized by Palevich et al. (Palevich et al.,
2019). The availability of a significant amount of MeHexAn-based oli­
gosaccharides following GH30 treatment further underlines that the
complexity of maize cell-wall polysaccharides transcends excessively

Due to the highly decorated nature of the cell-wall polysaccharides,
the efficiency of GH30, a xylanase active on xylans with MeGlcA sub­
stituents (Biely et al., 2016), was also analyzed. It was found that the
predominant PBOS produced by GH30 consisted of 8 or more pentose
units as opposed to the shorter chains observed for the GH5 type en­
zymes. This effect is very visible in Fig. 6, where the most intense signal
of Pen10-HexA-Ac comes from the area treated with GH30, while the
most intense signal of Pen6-HexA-Ac comes from the area treated with
GH5_34, a similar shift was also observed for Penn-Ac2-HexA (Fig. S4).
GH30 also showed a prevalence for producing more complex PBOS
9


J.R. Granborg et al.

Carbohydrate Polymers 275 (2022) 118693

branched xylans and a multienzyme approach is necessary to achieve
extensive degradation of these polysaccharides.

energy acquisition by xylanolytic Bacteroidetes*. Journal of Biological Chemistry, 285
(39), 30261–30273.
Ebringerov´
a, A. (2005). Structural diversity and application potential of hemicelluloses.
Macromolecular Symposia, 232(1), 1–12.

Ebringerov´
a, A., Hrom´
adkov´
a, Z., & Heinze, T. (2005). Hemicellulose. In T. Heinze (Ed.),
Polysaccharides I: Structure, characterization and use (pp. 1–67). Berlin, Heidelberg:
Springer Berlin Heidelberg.
Fanuel, M., Ropartz, D., Guillon, F., Saulnier, L., & Rogniaux, H. (2018). Distribution of
cell wall hemicelluloses in the wheat grain endosperm: A 3D perspective. Planta, 248
(6), 1505–1513.
Feenstra, A. D., Alexander, L. E., Song, Z., Korte, A. R., Yandeau-Nelson, M. D.,
Nikolau, B. J., & Lee, Y. J. (2017). Spatial mapping and profiling of metabolite
distributions during germination. Plant Physiology, 174(4), 25322548.
Gorzolka, K., Kă
olling, J., Nattkemper, T. W., & Niehaus, K. (2016). Spatio-temporal
metabolite profiling of the barley germination process by MALDI MS imaging. PLoS
One, 11(3), Article e0150208.
Grabber, J. H., Ralph, J., & Hatfield, R. D. (2000). Cross-linking of maize walls by
ferulate dimerization and incorporation into lignin. Journal of Agricultural and Food
Chemistry, 48(12), 6106–6113.
Huisman, M. M. H., Schols, H. A., & Voragen, A. G. J. (2000). Glucuronoarabinoxylans
from maize kernel cell walls are more complex than those from sorghum kernel cell
walls. Carbohydrate Polymers, 43(3), 269–279.
Kawamoto, T., & Kawamoto, K. (2021). Preparation of thin frozen sections from nonfixed
and undecalcified hard tissues using Kawamoto’s film method (2020). Methods in
Molecular Biology, 2230, 259–281.
Labourel, A., Crouch, L. I., Br´
as, J. L. A., Jackson, A., Rogowski, A., Gray, J., … Cuskin, F.
(2016). The mechanism by which arabinoxylanases can recognize highly decorated
xylans. Journal of Biological Chemistry, 291(42), 22149–22159.
Malunga, L. N., & Beta, T. (2016). Isolation and identification of feruloylated

arabinoxylan mono- and oligosaccharides from undigested and digested maize and
wheat. Heliyon, 2(5), Article e00106.
Ozturk, O. K., Kaasgaard, S. G., Palm´en, L. G., Vidal, B., & Hamaker, B. R. (2021). Protein
matrix retains most starch granules within corn fiber from corn wet-milling process.
Industrial Crops and Products, 165, Article 113429.
Palevich, N., Kelly, W. J., Ganesh, S., Rakonjac, J., Attwood, G. T., & Drake, H. L. (2019).
Butyrivibrio hungatei MB2003 competes effectively for soluble sugars released by
Butyrivibrio proteoclasticus B316T during growth on xylan or pectin. Applied and
Environmental Microbiology, 85(3) (e02056-02018).
P´
erez-Carrillo, E., & Serna-Saldívar, S. O. (2006). Cell wall degrading enzymes and
proteases improve starch yields of sorghum and maize. Starch - Stă
arke, 58(7),
338344.
Peukert, M., Lim, W. L., Seiffert, U., & Matros, A. (2016). Mass spectrometry imaging of
metabolites in barley grain tissues. Current Protocols in Plant Biology, 1(4), 574–591.
Puchart, V., Vrˇsansk´
a, M., Mastihubov´
a, M., Topakas, E., Vafiadi, C., Faulds, C. B., …
Biely, P. (2007). Substrate and positional specificity of feruloyl esterases for
monoferuloylated and monoacetylated 4-nitrophenyl glycosides. Journal of
Biotechnology, 127(2), 235–243.
Rausch, K. D., Hummel, D., Johnson, L. A., & May, J. B. (2019). Chapter 18 - Wet milling:
The basis for corn biorefineries. In S. O. Serna-Saldivar (Ed.), Corn (3rd ed., pp.
501–535). Oxford: AACC International Press.
Ravn, J. L., Glitsø, V., Pettersson, D., Ducatelle, R., Van Immerseel, F., & Pedersen, N. R.
(2018). Combined endo-β-1,4-xylanase and α-l-arabinofuranosidase increases
butyrate concentration during broiler cecal fermentation of maize glucuronoarabinoxylan. Animal Feed Science and Technology, 236, 159–169.
Robichaud, G., Garrard, K. P., Barry, J. A., & Muddiman, D. C. (2013). MSiReader: An
open-source interface to view and analyze high resolving power MS imaging files on

Matlab platform. Journal of the American Society for Mass Spectrometry, 24(5),
718–721.
Rosentrater, K. A., & Evers, A. D. (2018). Chapter 14 - wet milling: Separating starch,
gluten (protein) and fibre. In K. A. Rosentrater, & A. D. Evers (Eds.), Kent’s technology
of cereals (5th ed., pp. 839–860). Woodhead Publishing.
Rosicka-Kaczmarek, J., Komisarczyk, A., Nebesny, E., & Makowski, B. (2016). The
influence of arabinoxylans on the quality of grain industry products. European Food
Research and Technology, 242(3), 295–303.
Roth, C., Moroz, O. V., Turkenburg, J. P., Blagova, E., Waterman, J., Ariza, A., …
Wilson, K. S. (2019). Structural and functional characterization of three novel fungal
amylases with enhanced stability and pH tolerance. International Journal of Molecular
Sciences, 20(19), 4902.
Saulnier, L., Marot, C., Chanliaud, E., & Thibault, J.-F. (1995). Cell wall polysaccharide
interactions in maize bran. Carbohydrate Polymers, 26(4), 279–287.
Schramm, T., Hester, Z., Klinkert, I., Both, J.-P., Heeren, R. M. A., Brunelle, A.,
Ră
ompp, A. (2012). imzML — A common data format for the flexible exchange and
processing of mass spectrometry imaging data. Journal of Proteomics, 75(16),
5106–5110.
Stoeckli, M., Staab, D., Wetzel, M., & Brechbuehl, M. (2014). iMatrixSpray: A free and
open source sample preparation device for mass spectrometric imaging. Chimia
(Aarau), 68(3), 146–149.
Veliˇckovi´c, D., Saulnier, L., Lhomme, M., Damond, A., Guillon, F., & Rogniaux, H.
(2016). Mass spectrometric imaging of wheat (Triticum spp.) and barley (Hordeum
vulgare L.) cultivars: Distribution of major cell wall polysaccharides according to
their main structural features. Journal of Agricultural and Food Chemistry, 64(32),
6249–6256.

4. Conclusion
The use of MSI was successful for evaluation enzyme specific

degradation products of the arabinoxylan substrate in the endosperm of
maize kernels. It also showed the complex nature maize polysaccharides
through the variation of attached sidechains observed on the oligosac­
charide degradation products. Additionally, synergetic effects were
found when using combinations of different enzymes confirming that a
multi-enzyme solution would be necessary for sufficient degradation of
cell-wall polysaccharides in maize endosperm and thereby allowing an
increased release of starch.
CRediT authorship contribution statement
JRG and SGK conceived the study. JRG carried out experiments and
data analysis under the supervision of SGK and CJ. The manuscript was
written through contributions of all authors.
Acknowledgment
This work is partly funded by the Innovation Fund Denmark (IFD)
under File No. 8053-00212B. Support from the Carlsberg Foundation
and The Danish Council for Independent Research | Medical Sciences
(grant no. DFF – 4002-00391) is gratefully acknowledged.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.carbpol.2021.118693.
References
Agger, J., Viksø-Nielsen, A., & Meyer, A. S. (2010). Enzymatic xylose release from
pretreated corn bran arabinoxylan: Differential effects of deacetylation and
deferuloylation on insoluble and soluble substrate fractions. Journal of Agricultural
and Food Chemistry, 58(10), 6141–6148.
Appeldoorn, M. M., Kabel, M. A., Van Eylen, D., Gruppen, H., & Schols, H. A. (2010).
Characterization of oligomeric xylan structures from corn fiber resistant to
pretreatment and simultaneous saccharification and fermentation. Journal of
Agricultural and Food Chemistry, 58(21), 11294–11301.
Bajpai, P. (1997). Microbial xylanolytic enzyme system: Properties and applications. In

S. L. Neidleman, & A. I. Laskin (Eds.), Advances in applied microbiology (pp. 141–194).
Academic Press.
Bento-Silva, A., Vaz Patto, M. C., & do Ros´
ario Bronze, M. (2018). Relevance, structure
and analysis of ferulic acid in maize cell walls. Food Chemistry, 246, 360–378.
Biely, P., Singh, S., & Puchart, V. (2016). Towards enzymatic breakdown of complex
plant xylan structures: State of the art. Biotechnology Advances, 34(7), 1260–1274.
Bokhart, M. T., Nazari, M., Garrard, K. P., & Muddiman, D. C. (2018). MSiReader v1.0:
Evolving open-source mass spectrometry imaging software for targeted and
untargeted analyses. Journal of the American Society for Mass Spectrometry, 29(1),
8–16.
Chateigner-Boutin, A.-L., Ordaz-Ortiz, J. J., Alvarado, C., Bouchet, B., Durand, S.,
Verhertbruggen, Y., … Saulnier, L. (2016). Developing pericarp of maize: A model to
study arabinoxylan synthesis and feruloylation. Frontiers in Plant Science, 7(1476).
Chesson, A., Gordon, A. H., & Lomax, J. A. (1983). Substituent groups linked by alkalilabile bonds to arabinose and xylose residues of legume, grass and cereal straw cell
walls and their fate during digestion by rumen microorganisms. Journal of the Science
of Food and Agriculture, 34(12), 1330–1340.
Coelho, E., Rocha, M. A. M., Moreira, A. S. P., Domingues, M. R. M., & Coimbra, M. A.
(2016). Revisiting the structural features of arabinoxylans from brewers’ spent grain.
Carbohydrate Polymers, 139, 167–176.
de Lima, E. A., Machado, C. B., Zanphorlin, L. M., Ward, R. J., Sato, H. H., & Ruller, R.
(2016). GH53 Endo-Beta-1,4-Galactanase from a newly isolated bacillus
licheniformis CBMAI 1609 as an enzymatic cocktail supplement for biomass
saccharification. Applied Biochemistry and Biotechnology, 179(3), 415–426.
de O. Buanafina, M. M. (2009). Feruloylation in grasses: Current and future perspectives.
Molecular Plant, 2(5), 861–872.
Dodd, D., Moon, Y.-H., Swaminathan, K., Mackie, R. I., & Cann, I. K. O. (2010).
Transcriptomic analyses of xylan degradation by Prevotella bryantii and insights into

10