Analytical Biochemistry 597 (2020) 113668

Contents lists available at ScienceDirect

Analytical Biochemistry
journal homepage: www.elsevier.com/locate/yabio

Identifying transglutaminase reaction products via mass spectrometry as T

exemplified by the MUC2 mucin - Pitfalls and traps☆
Liisa Arike, Gunnar C. Hansson, Christian V. Recktenwald∗
Department of Medical Biochemistry, University of Gothenburg, Gothenburg, Sweden

A R T I C LE I N FO A B S T R A C T

Keywords: In order to demonstrate transglutaminase activity in biological samples immunological as well as glutamine- and
Mucus amine-donor based assays are commonly used. However, the identification of the transglutaminase reaction
MUC2 product, i. e. the isopeptide cross-linked peptides/proteins or the deamidated protein/peptide are often ne-
Isopeptide bond glected. This article describes a workflow for the detection of the products of transglutaminase-catalyzed re-
Cross-link
actions. In particular, possible pitfalls and traps that can arise during the mass spectrometry-based identification
Transglutaminase
of isopeptide cross-links are addressed and characterised on actual samples.

Transglutaminases (TGases, E.C. 2.3.2.13) are a family of amino
acyl-transferases that can catalyze the Ca2+-dependent formation of a
covalent bond between the side chains of protein-bound glutamine and
lysine accompanied by the release of ammonia. The nature of the en-
zyme-substrate intermediate complex is a thioester bond, to which
nucleophiles like primary amines such as the ε-amino group of lysine or
water can perform a nucleophilic attack thereby producing either an
isopeptide bond or a glutamic acid residue [1]. TGase activity in cel-
lular extracts is typically detected by the incorporation of either a
biotinylated acyl-donor or an acceptor peptide, and subsequent gel
electrophoresis and Western blot [2]. Some researchers have demon-
strated TGase activity in tissue sections by using an antibody against the
TGase reaction product and co-immunostaining against the cross-linked
target protein [3]. Despite the usefulness of these approaches they fail
to localize the respective glutamine and lysine residue of a TGase-cat-
alyzed isopeptide cross-link. Hence, an essential piece of information is
missing since the introduction of an isopeptide bond results in im-
portant structural and functional changes of the target protein(s) as it
has been shown for the self-multimerization of tissue transglutaminase
[4]. In order to map the residues forming the isopeptide bond mass
spectrometric (MS) analysis has become the method of choice. Several
studies have utilized an in vitro approach where synthesized gliadin
peptides have been treated with TGase 2 and the cross-linked TG2-
gliadin complexes were subsequently identified by MS analyses [5–7].
Other studies for the detection of protein substrates for TGases used a
strategy whereby labelled acyl-donor or –acceptor compounds were
added to separate reaction mixtures of e.g. a cell or tissue lysate, and
the modified residues were later identified by a database search [8–10].
All these approaches have been proven to be very useful for the iden-
tification of new putative TGase substrates and the identification of
preferred substrate residues. However, they have two major drawbacks.
Firstly, it is not possible to identify the “natural” cross-linked protein
substrates inside a cell or biological fluid. Second, TGases are pro-
miscuous when it comes to the amine-donor. Therefore, any freely ex-
posed lysine residue or a non-acetylated protein N-terminus of the
protein surface can perform the nucleophilic attack and form the
covalent bond with the glutamine-donor peptide. Due to the small size
of glutamine-donor peptides the influence of steric hindrance when
bound to the catalytic cysteine of the enzyme is reduced when com-
pared to a protein substrate, thereby increasing the chance of false-
positive identifications.
To overcome these limitations it is necessary to isolate the TGase
reaction products from the biological sample before MS analysis. This
article describes the workflow for the analysis of cross-linked proteins
on the gel-forming mucin MUC2 as an example and points to pitfalls
that can arise during this process. Furthermore, possible traps for the
analysis of deamidated TGase reaction products are addressed as well.
The gel-forming mucin MUC2 is the major constituent of the

☆
This work was supported by the European Research Council ERC (694181), National Institute of Allergy and Infectious Diseases (U01AI095473, the content is
solely the responsibility of the authors and does not necessarily represent the official views of the NIH), The Knut and Alice Wallenberg Foundation (2017.0028),
Swedish Research Council (2017-00958), The Swedish Cancer Foundation (CAN 2017/360), IngaBritt and Arne Lundberg Foundation, Sahlgren's University Hospital
(ALFGBG-440741, The ALF agreement 236501), Bill and Melinda Gates Foundation (OPP1202459), Wilhelm and Martina Lundgren’s Foundation.
∗
Corresponding author.
E-mail address: christian.recktenwald@medkem.gu.se (C.V. Recktenwald).

https://doi.org/10.1016/j.ab.2020.113668
Received 30 November 2019; Received in revised form 15 February 2020; Accepted 27 February 2020
Available online 25 March 2020
0003-2697/ © 2020 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
L. Arike, et al. Analytical Biochemistry 597 (2020) 113668

Fig. 1. Flowchart of the analytical workflow for the mass spectrometric identification of transglutaminase-catalyzed isopeptide cross-links. Panel A) summarizes a
general workflow for the identification of naturally occurring TGase reaction products in a biological sample. Panel B) shows the protocol for the detection of TGase-
mediated MUC2-oligomers. The included gel scan shows the composite agarose-PAGE-separated MUC2 monomer and the respective multimers from mouse colonic
mucus visualized with Alcian Blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

intestinal mucus that covers the epithelial surface thereby protecting it
from bacterial induced inflammation. MUC2 oligomerises via disulfides
involving both the N- and C-terminus. During the later stages of its
biosynthesis in the trans-Golgi network reduction insensitive bonds
between MUC2 monomers are introduced, that have been shown to be
isopeptide bonds catalyzed by TGases [11].
One major obstacle for the detection of TGase-mediated cross-linked
proteins is that they are far less abundant than linear peptides in a
biological sample. Therefore, traditional shotgun proteomics ap-
proaches are insufficient as there are too many co-eluting peptides
during the LC separation that will suppress the ionization of the cross-
linked peptides. To overcome this, it is necessary to enrich for the TGase
reaction products via the most suitable method(s), e. g. size exclusion
chromatography, SDS-PAGE or affinity purification (Fig. 1A).
In the case of the MUC2 mucin, it is possible to take advantage of its
insolubility in 6 M guanidinium chloride (GuHCl) and its molecular
weight (Fig. 1B). By using repeated extraction steps in GuHCl most
other proteins will be washed away. After this MUC2 can be brought
into solution by the addition of dithiothreitol (DTT) and it is subse-
quently alkylated with iodoacetamide (IAA). The molecular mass of
MUC2 is 650 kDa for the primary translation product. After extensive
N- and O-glycosylation the mass of the MUC2 monomer increases to
approximately 2.5 MDa. Therefore, it is possible to obtain almost pure
TGase–based oligomers after separation by composite agarose-poly-
acrylamide gel electrophoresis (AgPAGE) [12]. After staining of the
MUC2 monomer and its multimers with Alcian Blue the bands in the gel
are subjected to protease digestion before MS analysis.
As the obtained peptides can become large the choice of the pro-
tease can be critical for the outcome of the experiment. This is espe-
cially important for the analysis of cross-links where lysine residues are
part of the covalent bridge as in TGase-catalyzed isopeptide bonds be-
cause the commonly used protease trypsin will not cleave once the
lysine residue has been modified [13]. Therefore, the addition of a
second protease or choosing a protease with different specificity is
beneficial. Chymotrypsin generally produces shorter peptides than
trypsin. However, for MS analyses chymotrypsin should be avoided
because the new C-termini are aromatic amino acids or leucine which
do not easily take up protons, resulting in the suppression of the ioni-
zation efficiency. For the digestion of the MUC2 mucin a combination of
trypsin and AspN is recommended. The protein bands of the composite
gel are first digested with trypsin followed by AspN after blocking the
trypsin activity with phenylmethanesulfonyl fluoride. Afterwards, the
samples are cleaned from salts and buffer compounds by C18 solid
phase extraction and separated on RP-nLC before MS analysis. Since the
sample complexity increases by n2 for cross-linked peptides high re-
solving power and mass accuracy become very important for the correct
identification of deamidated and transamidated TGase-catalyzed reac-
tion products. We suggest a resolution of at least 60,000 for the parent
ions and 15,000 at the MS2 level. During the data analysis the number
of false-positive identifications should be limited by setting the mass
accuracy to 1–2 ppm for the parent ions and to 10–20 ppm for fragment
ions.
There are several software packages available for the analysis of
cross-linked peptides e.g. Kojak, pLink, ProteinProspector, StavroX

2
L. Arike, et al.
[14–17]. In order to detect cross-linked peptides with these softwares
the user has to define its database, the proteolysis of the proteins, static
and variable modifications, error tolerances for MS1 and MS2 level as
for regular MS database search. In addition, the cross-linked amino acid
residues, i. e. Gln and Lys, protein N-terminus for TGase reaction pro-
ducts and the introduced composition/mass (shift) of the cross-linker, i.
e. –NH3/-17.03 for TGase-catalyzed isopeptide-bonded dipeptides, have
to be specified for the detection of the cross-linked products. Therefore,
3
Analytical Biochemistry 597 (2020) 113668
Fig. 2. Examples of identified isopeptide cross-
links from an MUC2 oligomer. A) MS2 fragment
spectrum of the parent ion [M+4H]4+ 717.81,
annotated as isopeptide cross-link after analysis
by the StavroX software and manually inserted
into the raw spectrum. B) MS2 fragment spec-
trum of the same parent ion [M+4H]4+ 717.81,
annotated as deamidated version using the
ProteinProspector search engine. Both annota-
tions cover the same consecutive sequence
INKPEVQCEDPEAVQEPESCSEHR corresponding
to amino acid residues 197–220 from murine
-Muc2. The deamidated version in B) contains a
missed cleavage site for AspN N-terminal of as-
partate 206. C) MS2 fragment spectrum of the
parent ion [M+3H]3+ 517.30. Detection of the
b1 ion excludes the formation of pyro-glutamate
at the N-terminal glutamine. Fragment ions la-
belled in brackets (y7 to y9) suggest a possible
break of the isopeptide bond in the mass spec-
trometer. [II] and [YI] mark internal fragment
ions. Underlined cysteines correspond to carba-
midomethylation of the residue, underlined
glutamine corresponds to a deamidation.
the identification of the TGase-mediated isopeptide cross-link relies on
the alteration of the dipeptide mass by the loss of ammonia. Since a
cross-link that occurs between two peptides (type 2, inter-link) contains
one additional N- and C-terminus in comparison to the same peptide
sequence as a linear peptide the net mass gain is one mass unit for
TGase reaction products. This raises specific problems for the identifi-
cation of TGase-mediated cross-links in consecutive peptide sequences
containing a protease cleavage site. The deamidated form of this
L. Arike, et al.
peptide with a missed cleavage will have the same mass as the cleaved
and transamidated dipeptide of the same sequence. In fact, they are
isomers as the molecule composition changes by -N-H+O for both
modifications and therefore have the same parent ion mass. In order to
verify if the consecutive peptide sequence has been modified by an
isopeptide bond or by a deamidation a regular database search with
deamidation of glutamine/asparagine should be performed in parallel.
If the same parent ion with an assigned deamidation is detected in this
search the two MS2 spectra should be revisited and compared for in-
dicator ions. However, in the absence of indicative ions for one or the
other modification it will not be possible to correctly annotate the
modification. Fig. 2A and B show an example of such a case from the
analysis of mouse Muc2 oligomers where the same precursor was
identified as cross-linked and deamidated peptide. Due to the possible
annotation of the peaks at m/z 1258, 1357 and 1428 as y10, y11 and
y12 fragments of the deamidated peptide sequence this modification
could be assigned (Fig. 2B). On the other hand, the same spectrum
contains the b2 to b4 fragments of the α-peptide for the isopeptide
cross-linked dipeptide what could argue for this modification (Fig. 2A).
Hence, it is not possible to determine which of the two versions is the
correct one. This result might be explained by the known observation
that TGase-mediated transamidation and deamidation can happen in
parallel [18]. It is expected that a linear deamidated peptide with
missed cleavage site and the transamidated dipeptide version would
elute at different retention times. However, in this particular case the
spectra for both products were identified at the same retention time
resulting in a mixed MS2 spectrum from two isomeric precursors which
prevents a distinct decision for one or the other. This could suggest that
both modifications were obtained by TGase activity and exist in par-
allel. Therefore, indications for the cross-linked product should be
taken from orthogonal data like identification of the cross-linked pep-
tide in the dimeric form of the protein, e. g. after an initial separation
on a polyacrylamide gel.
Another modification that can lead to false-positive identifications
of the transamidated reaction product is the formation of pyro-gluta-
mate (pyroGlu). This post-translational modification occurs during LC-
MS analysis if the N-terminus of a peptide consists of glutamine or
glutamate [19]. For glutamine this results in the same loss of ammonia
as for the TGase-catalyzed intra-molecular cross-link in this peptide
(type 1, loop). Hence, inspection of the fragment spectrum is necessary
to verify the TGase-catalyzed cross-link. Fig. 2C shows an example of
such a peptide where pyroGlu formation can be excluded due to the
presence of the b1-ion. In this spectrum the presence of the y7 to y9
ions suggest that the isopeptide bond has been broken in the mass
spectrometer as has been shown by the work of Kang and Baker [20].
Since this is not a well-studied fragmentation event the respective
fragments are labelled in brackets. However, in spectra where pyroGlu
formation cannot be excluded the possibility of a cross-link should be
discarded.
Finally, in order to quantify and validate the data we suggest a
targeted MS method, parallel reaction monitoring (PRM) together with
the open source software package Skyline. The PRM method does not
require any previous knowledge of fragment ions, therefore it is easy to
apply and it improves the spectral quality for low abundant and com-
plex cross-linked peptides. Skyline [21] adds automation of the quan-
tification into the pipeline, which is beneficial when larger datasets
have to be analysed. The combination of PRM analysis and Skyline has
been applied for chemically cross-linked peptides and protocols for such
workflow exist [22,23]. As Skyline is not designed to analyse naturally
occurring cross-linked peptides two different approaches can be used
depending on which kind of cross-linked peptides are analysed. For
TGase-mediated isopeptide bonds in the same peptide, type 1, loop
cross-link, loss of NH2 on glutamine and loss of H on lysine are suffi-
cient in order to insert the crosslinked peptide into Skyline. For type 2,
inter-linked peptides, the identified cross-link needs to be added as a
small molecule and the fragment ions will be inserted manually, based
4
Analytical Biochemistry 597 (2020) 113668
on the previous knowledge from the search engine.
In comparison to the detection of TGase-catalyzed transamidated
peptides the identification of deamidated reaction products is more
straightforward as most search engines provide this post-translational
modification as variable modification. However, the identification of
deamidated TGase reaction products has also several problems.
Therefore, care has to be taken during data acquisition and peak
picking. The mass of the first 13C peak and the deamidation product
differ only by 19 m mass units for charge state one. This can cause
wrong assignments of overlapping peaks from the amidated form of the
peptide [24]. Hence, the distinction between the two peaks requires
high resolving power and the mass spectrometer should be set to a
resolving power of at least 60,000. In addition, deamidation can occur
spontaneously [25], requiring appropriate controls to be analysed in
parallel to the actual sample. This can easiest be achieved by treatment
of a sample aliquot with a TGase inhibitor.
In conclusion, the advances in high resolution accurate mass spec-
trometry make it possible to identify and map glutamine- and lysine-
residues involved in TGase-catalyzed transamidation and deamidation
reactions. This information on the localization of these sites can provide
important information about structural and functional changes in the
TGase substrate proteins. For example, identification of the isopeptide
cross-links occurring in vivo will be useful to answer questions of homo-
or hetero-oligomerisation in target proteins and the composition of
protein networks. However, the analysis of isopeptide cross-links from
biological sources remains challenging due to their low abundance and
the potential of false-positive results due to ammonia loss during TGase-
mediated cross-linking. Therefore, careful evaluation and appropriate
controls are needed in order to verify the obtained results.
Acknowledgements
We thank Brendan Dolan for helpful comments on the manuscript
and Sjoerd van der Post for helpful discussions.
References
[1] L. Lorand, R.M. Graham, Transglutaminases: crosslinking enzymes with pleiotropic
functions, Nat. Rev. Mol. Cell Biol. 4 (2) (2003) 140–156.
[2] S. Altuntas, et al., The transglutaminase type 2 and pyruvate kinase isoenzyme M2
interplay in autophagy regulation, Oncotarget 6 (42) (2015) 44941–44954.
[3] M. de Jager, et al., Tissue transglutaminase colocalizes with extracellular matrix
proteins in cerebral amyloid angiopathy, Neurobiol. Aging 34 (4) (2013)
1159–1169.
[4] J. Stamnaes, et al., Enhanced B-cell receptor recognition of the autoantigen trans-
glutaminase 2 by efficient catalytic self-multimerization, PloS One 10 (8) (2015)
e0134922.
[5] B. Fleckenstein, et al., Molecular characterization of covalent complexes between
tissue transglutaminase and gliadin peptides, J. Biol. Chem. 279 (17) (2004)
17607–17616.
[6] S. Dorum, et al., The preferred substrates for transglutaminase 2 in a complex wheat
gluten digest are Peptide fragments harboring celiac disease T-cell epitopes, PloS
One 5 (11) (2010) e14056.
[7] B. Lexhaller, C. Ludwig, K.A. Scherf, Comprehensive detection of isopeptides be-
tween human tissue transglutaminase and gluten peptides, Nutrients 11 (10)
(2019).
[8] Y. Ito, et al., Detection and identification of potential transglutaminase 2 substrates
in the mouse renal glomeruli, Arch. Biochem. Biophys. 660 (2018) 11–19.
[9] K.N. Lee, et al., Identification of transglutaminase substrates in HT29 colon cancer
cells: use of 5-(biotinamido)pentylamine as a transglutaminase-specific probe,
Biochim. Biophys. Acta 1136 (1) (1992) 12–16.
[10] H. Tatsukawa, et al., Global identification and analysis of isozyme-specific possible
substrates crosslinked by transglutaminases using substrate peptides in mouse liver
fibrosis, Sci. Rep. 7 (2017) 45049.
[11] C.V. Recktenwald, G.C. Hansson, The reduction-insensitive bonds of the MUC2
mucin are isopeptide bonds, J. Biol. Chem. 291 (26) (2016) 13580–13590.
[12] B.L. Schulz, N.H. Packer, N.G. Karlsson, Small-scale analysis of O-linked oligo-
saccharides from glycoproteins and mucins separated by gel electrophoresis, Anal.
Chem. 74 (23) (2002) 6088–6097.
[13] J. Baeza, et al., Stoichiometry of site-specific lysine acetylation in an entire pro-
teome, J. Biol. Chem. 289 (31) (2014) 21326–21338.
[14] M.R. Hoopmann, et al., Kojak: efficient analysis of chemically cross-linked protein
complexes, J. Proteome Res. 14 (5) (2015) 2190–2198.
[15] B. Yang, et al., Identification of cross-linked peptides from complex samples, Nat.
L. Arike, et al.
Methods 9 (9) (2012) 904–906.
[16] M.J. Trnka, et al., Matching cross-linked peptide spectra: only as good as the worse
identification, Mol. Cell. Proteomics 13 (2) (2014) 420–434.
[17] M. Gotze, et al., StavroX–a software for analyzing crosslinked products in protein
interaction studies, J. Am. Soc. Mass Spectrom. 23 (1) (2012) 76–87.
[18] J. Stamnaes, B. Fleckenstein, L.M. Sollid, The propensity for deamidation and
transamidation of peptides by transglutaminase 2 is dependent on substrate affinity
and reaction conditions, Biochim. Biophys. Acta 1784 (11) (2008) 1804–1811.
[19] P. Purwaha, et al., An artifact in LC-MS/MS measurement of glutamine and glu-
tamic acid: in-source cyclization to pyroglutamic acid, Anal. Chem. 86 (12) (2014)
5633–5637.
[20] H.J. Kang, E.N. Baker, Intramolecular isopeptide bonds: protein crosslinks built for
stress? Trends Biochem. Sci. 36 (4) (2011) 229–237.
Analytical Biochemistry 597 (2020) 113668
[21] B. MacLean, et al., Skyline: an open source document editor for creating and ana-
lyzing targeted proteomics experiments, Bioinformatics 26 (7) (2010) 966–968.
[22] J.D. Chavez, et al., A general method for targeted quantitative cross-linking mass
spectrometry, PloS One 11 (12) (2016) e0167547.
[23] Z.A. Chen, J. Rappsilber, Protein dynamics in solution by quantitative crosslinking/
mass spectrometry, Trends Biochem. Sci. 43 (11) (2018) 908–920.
[24] A.I. Nepomuceno, et al., Accurate identification of deamidated peptides in global
proteomics using a quadrupole orbitrap mass spectrometer, J. Proteome Res. 13 (2)
(2014) 777–785.
[25] A. Serra, et al., Characterization of glutamine deamidation by long-length electro-
static repulsion-hydrophilic interaction chromatography-tandem mass spectrometry
(LERLIC-MS/MS) in shotgun proteomics, Anal. Chem. 88 (21) (2016)
10573–10582.