Vaccine 25 (2007) 2213–2227
Investigation of the detoxification mechanism of
formaldehyde-treated tetanus toxin
Morten Thaysen-Andersen a , Sys Borcher Jørgensen b , Ellen Sloth Wilhelmsen b ,
Jesper Westphal Petersen b , Peter Højrup a,∗
a
Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark
b Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen S, Denmark
Received 19 May 2006; received in revised form 13 October 2006; accepted 7 December 2006
Available online 2 January 2007
Abstract
The tetanus vaccine is based on the extremely potent tetanus neurotoxin (TeNT), which is converted by treatment with formaldehyde and
lysine into the non-toxic, but still immunogenic tetanus toxoid (TTd). This formaldehyde-induced detoxification, which to a large extend
determines the quality and properties of the vaccine component, occurs through partly unknown chemical modifications of the toxin.
The aim of this study was to gain knowledge of the detoxification mechanism in the generation of the tetanus vaccine. Two approaches
were chosen: (i) the effect of changes in the concentrations of lysine and formaldehyde in the detoxification process and (ii) characterisation
of the chemically detoxified TTd.
(i) We examined a number of TTd components that was produced by varying the concentrations of formaldehyde and lysine during
the inactivation. Toxicity tests showed that the detoxification failed when the lysine or formaldehyde concentration was ≤1/5 or ≤1/10,
respectively, of the standard level. Gel-electrophoretic analyses showed that inter-chain cross-linking was formaldehyde-dependent and,
furthermore, revealed that inter-chain cross-linking was not the only requirement for the inactivation. In addition, the measurable amount of
tyrosine correlated inversely with the degree of inter-chain cross-linking.
(ii) To study the formaldehyde-induced chemical modifications, the TTd was investigated using protein chemical techniques in combination with mass spectrometry (MS). Using off-line liquid chromatography (LC)–MS, the most pronounced chemical modifications were
characterised as unstable Schiff-bases (+12 Da) located on lysine residues and the N-termini of peptides throughout the molecule. Several
arginine residues were also found with +12 Da modifications due to Schiff-base formation or as a consequence of degenerative fragmentation
of lysine/formaldehyde adducts or cross-links during MS. A few tyrosine residues were similarly observed with a mass increase of 12 Da.
Even though it cannot be ruled out that this is a residual mass of higher molecule adducts or cross-links to tyrosine, amino acid analysis and MS
data indicated that the modification forms a ring structure from a carbon in the aromatic ring to the backbone N␣ . In addition, several mono␧-methyllysines (+14 Da) were observed as a likely consequence of reductive methylation of the Schiff-bases. A substantial part (87%) of the
known TeNT sequence, including the active site, was covered using the off-line LC–MS approach to investigate the tryptic digested TTd. In
contrast to the results obtained from the gel-electrophoretic experiments, neither intra/inter-chain cross-links nor cross-links to external lysines
were observed in the MS analysis. Instability of the cross-links during separation and/or MS is likely to explain their absence in the analyses.
The biological relevance of the observed modifications is discussed in relation to 3D mapping analyses. Proposals for the TeNT detoxification
are discussed, although no direct evidence for the exact mechanism could be obtained.
© 2006 Elsevier Ltd. All rights reserved.
Keywords: Tetanus toxoid; Formaldehyde; Mass spectrometry
Abbreviations: ESI, electrospray ionisation; FA, formic acid; MALDI, matrix-assisted laser desorption/ionisation; MeCN, acetonitrile; MS, mass spectrometry; MS/MS, tandem mass spectrometry; Q, quadrupole; RP-HPLC, reversed-phase high performance liquid chromatography; SDS-PAGE, sodium dodecyl
sulfate-polyacrylamide gel electrophoresis; SPITC, 4-sulfophenyl isothiocyanate; TeNT, tetanus neurotoxin; TFA, trifluoroacetic acid; TOF, time-of-flight; TTd,
tetanus toxoid
∗ Corresponding author. Tel.: +45 6550 2371; fax: +45 6550 2467.
E-mail address: php@bmb.sdu.dk (P. Højrup).
0264-410X/$ – see front matter © 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.vaccine.2006.12.033
2214
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
1. Introduction
Clostridium tetani, the bacterium responsible for the
production of the extremely potent tetanus neurotoxin
(TeNT) leading to tetanus, is widespread in nature in the
form of spores. Under certain conditions of very low oxygen
tension, slight acidity and availability of nutrients, such as
anaerobic wounds or skin ruptures, the spores can germinate
and produce the TeNT [1]. The mature TeNT contains a
50 kDa light chain disulfide bonded to a 100 kDa heavy
chain [2,3].
In 1924 Ramon demonstrated that toxins could be inactivated by formaldehyde treatment, producing the non-toxic
toxoid [4]. Importantly, the formaldehyde-induced detoxification of the toxins does not destroy the immunogenic sites
of the molecules, retaining the antibody-producing abilities.
The detoxification process is very important in determining the quality of the tetanus toxoid (TTd) as a vaccine
component. Here, reaction conditions such as formaldehyde
concentration, reaction time, temperature and concentration
of the reaction matrix (glycine or lysine solutions) are critical
parameters.
Formaldehyde is, together with other aldehydes, a wellknown cross-linking agent and is widely used in various areas
of research. A number of studies have been performed on the
reactions of formaldehyde with mixtures of amino acids and
small peptides in order to determine the nature of the reaction [5–9]. It is generally accepted that formaldehyde can
react with primary amino groups to form methylol derivatives, and that this group can undergo condensation to a
Schiff-base (Fig. 1). Subsequently, the Schiff-base can form
cross-links with several amino acid residues, including lysine
and tyrosine. It is also known that the Schiff-base can undergo
reductive methylation in the presence of a reducing agent, e.g.
NaCNBH3 .
Some aspects of the cross-linking reaction remain
poorly understood in particular when dealing with larger
molecules where higher levels of structures affect the
chemistry. Besides the local environment, the pH, the rate
of a particular cross-link reaction, the components present
in the reaction solution and the reactant concentrations
are factors, that are known to influence the formation of
modifications. Recently, Metz et al. performed a study on
formaldehyde-induced modifications in a protein [10]. Even
though the study was carried out on a small protein (insulin),
this is to our knowledge, the first investigation dealing
with the formaldehyde chemistry in structures of higher
levels.
Despite the fact that most international regulations gradually allow the shift towards the use of alternative test methods
for the quality control of vaccines [11], the present methods for determining the quality of tetanus and diphtheria still
depend on animal testing [12–14]. To date, no in vitro test
has been accepted as an alternative method for the quality
determination of these toxoids. Another concept of quality
control is based on a highly consistent production process,
where the vaccine batches have identical properties. This way,
the potency of a newly produced toxoid can be predicted, if
the new product is indistinguishable from a reference toxoid with a proven potency [13–15]. This is now common
practice for some well-defined biologicals [16]. A number
of physicochemical properties of the molecules are parameters that allow investigation of the vaccine components, e.g.
size, structure, purity, amino acid modifications and HPLCprofile. Combining these results may be sufficient for proving
that vaccines have identical properties and are consistently
produced.
In order to implement these new in vitro tests, it is essential to obtain knowledge about the detoxification mechanism
of formaldehyde-treated toxins as well as to characterise the
structural changes of the molecule during this treatment. For
this purpose we monitored the tetanus detoxification process
using various detoxification conditions and performed toxicity tests and structural analyses on the resulting components
Fig. 1. Formaldehyde chemistry. (i) The primary amino group reacts with formaldehyde to form a methylol derivative of +30 Da. This product can condensate
to a Schiff-base (+12 Da). In the presence of a reducing agent the Schiff-base is reduced to a methyl group (+14 Da). (ii) The Schiff-base can also react with
several amino acid residues, e.g. to form cross-links through methylene bridges.
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
using a number of methods including amino acid analysis,
SDS-PAGE, gel-filtration, reversed-phase (RP)-HPLC and
mass spectrometry.
2215
Table 1
(a) Design of matrix experimentsa and (b and c) the relative quantities (in
percentage) of tyrosine and lysineb
2. Materials and methods
2.1. Production and purification of TTd
TeNT was purified from fermentation cultures of C.
tetani before detoxification. Crude TeNT was harvested,
dialysed, filtrated and precipitated by ammoniumsulfate.
After re-solubilization, TeNT was dialysed and filtrated and
TeNT concentration, purity and pH were measured. The
purified TeNT was inactivated at a fixed concentration of
500 ± 50 Lf/ml by adding formaldehyde, lysine and phosphate buffer to final concentrations of 27, 10 and 27 mM,
respectively. The mixture was adjusted to pH 7.4 ± 0.2 and
incubated at 35 ± 2 ◦ C for 4 weeks. The pH and toxin/toxoid
concentration was monitored each week and, if necessary,
pH was adjusted to fit into the specified pH range. The
resulting TTd was then dialysed and filtrated to remove the
excess inactivation reagents. After a final pH adjustment
and filtration, the TTd was aliquoted in appropriate volumes
and stored as bulk TTd. The TTd was analysed with a set
of final product release tests, estimating the quality of the
product.
2.2. Matrix experiments
In order to investigate the effect of formaldehyde and
lysine in the detoxification, TeNT was detoxified using
different conditions than described above. Varying the
formaldehyde concentration to four levels (54 and 27 mM
(standard concentration), 13.5 and 6.75 mM) and the lysine
concentration to four levels (20 and 10 mM (standard concentration), 5 and 2.5 mM), a 4 × 4 matrix of samples was
set up (TTd01–TTd16). In another experiment the lysine
and formaldehyde concentrations were varied to three levels to produce a 3 × 3 matrix (TTd06 and TTd17–TTd24).
Here, the formaldehyde concentrations were varied to 27,
2.7 and 0.27 mM and the lysine concentrations varied to
10, 2 and 0.4 mM. Table 1a illustrates the design of the
experiments.
All conditions other than the concentrations of lysine and
formaldehyde were identical to the production of the commercial tetanus vaccine.
2.3. Toxicity tests
The toxicity of the matrix samples (TTd01–TTd24) were
determined by injecting 0.5 ml of each matrix sample diluted
(1:2) in saline subcutaneously into a group of five mice
(female CD1 mice (16–25 g)). The mice were observed for 14
days for any signs of tetanus paralysis. Mice showing tetanus
symptoms were sacrificed immediately.
a Two critical parameters, the formaldehyde and lysine concentration,
were varied in the detoxification process relative to the standard concentration ‘1’. The samples treated with the most reduced concentrations (3 × 3
matrix including TTd06, light gray) induced tetanus symptoms in mice
(marked with ‘T’), whereas the other samples (4 × 4 matrix, dark gray)
appeared non-toxic. Samples are labelled as indicated (TTd01–TTd24).
b The relative quantities (in percentage) of tyrosine and lysine of
TTd01–TTd24, TeNT and TTd are listed as average values based on duplicate
measurements omitting proline, tryptophan and cysteine.
2.4. SDS-PAGE of tetanus toxin, toxoid and matrix
samples
TeNT, TTd and TTd01–TTd24 (∼5 ␮g each) were mixed
with 5 ␮l of reducing sample buffer containing 5% ␤mercaptoethanol and loaded on a 4–20% Tris–Glycine
PAGEr Gold Precast Gel (Cambrex, Rockland, ME, USA).
One hundred and sixty volts were applied to the gel. The gel
was subsequently stained with Coomassie blue (dissolved
in fixing solution containing 45% methanol and 10% acetic
acid in water) and later destained using water. The water used
for analytical experiments was of ultra-high quality (Purelab
Ultra, Vivendi Water Systems, USA).
2.5. Amino acid analysis
Samples were dried in 500 ␮l polypropylene vials. The
lids were punctured and the vials placed in 25 ml glass vials
equipped with MinInert valves (Pierce Biotechnology, Rockford, IL, USA). One hundred to two hundred microliters of
2216
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
6N HCl containing 0.1% phenol was placed in the bottom of
the glass and blown with argon before a vacuum was applied.
The samples were incubated at 110 ◦ C for 18 h. The samples
were subsequently redissolved in 50 ␮l 0.20 M sodium citrate
loading buffer pH 2.20 (Biochrom, Cambridge, UK), transferred to microvials and loaded on a BioChrom 30 amino
acid analyzer (Biochrom). Data analysis was performed using
in-house developed software.
Gelfiltration of a tryptic digest of TTd was performed
using a Superdex peptide PC 3.2/30 column (Amersham
Pharmacia Biotech). Prior to analysis, the sample was dried to
a small volume (∼10 ␮l). The mobile phase consisted of 30%
MeCN, 0.1% TFA in water and the flow was kept constant at
50 ␮l/min.
2.6. Edman degradation
When possible, peptides were applied directly onto
the MALDI target plate using the dried droplet-method
without prior micropurification [17]. In the dried-droplet
method 0.5 ␮l of 2% TFA was loaded on the target and
added 1 ␮l of analyte before an additional 0.8 ␮l of matrix
solution was added and the sample was dried. ␣-Cyano-4hydroxycinnamic acid, dissolved in 0.1% TFA, 70% MeCN
at a concentration of 10 mg/ml, was used as matrix.
Samples containing salt or analytes of low concentrations were micropurified using hydrophobic column material
in combination with target load. GelLoader pipette tips
(Eppendorf, Hamburg, Germany) were partially constricted
by squeezing the end and 5 ␮l slurry (Poros R2, 20 ␮m,
Applied Biosystems, Framingham, MA, USA) was loaded
onto the tip and packed by applying air pressure with a 1 ml
syringe. The resin (1–5 mm in height) was equilibrated by
flushing 10 ␮l 5% FA through the column. The sample was
loaded onto the column in an appropriate hydrophilic buffer
and washed by another 10 ␮l of 5% FA. The sample was
eluted onto the target by 0.8 ␮l of matrix solution.
Some peptides were N-terminally derivatised prior to
MALDI TOF MS2 . Here, RP-HPLC separated peptides
(5–10 pmol) were mixed with 8.5 ␮l of reagent solution consisting of 10 ␮g/␮l 4-sulfophenyl isothiocyanate (SPITC)
(Sigma–Aldrich, St. Louis, MO, USA) dissolved in 50 mM
NaHCO3 , pH 8.6. The mixture was incubated at 56 ◦ C for
30 min before 1 ␮l 5% TFA was added to terminate the
reaction. Subsequently, the derivatised peptides were micropurified as described above prior to MS.
Approximately 200 pmol of intact TeNT and TTd were
applied to biobrene-treated glass filters and run for 15 cycles
on an Applied Biosystems 494 protein sequencer (Foster City,
CA, USA) essentially as described by the manufacturer.
2.7. Proteolytic digests
Two nanomoles (∼300 ␮g) of sample (TeNT, TTd and
TTd01–TTd24) was dried and redissolved in 50 ␮l 0.2%
(w/v) Rapigest SF (Waters, Milford, MA, USA) in 100 mM
NH4 HCO3 . Fifteen microliters 50 mM 1,4-dithiothreitol
in water was added and the mixture was incubated at
56 ◦ C for 30 min and subsequently chilled to R.T. Fifteen
microliters 100 mM iodoacetamide in water was added and
the mixture incubated for 30 min at R.T. in the dark. Twenty
micrograms of modified trypsin (Promega, Madison, WI,
USA) was dissolved in 20 ␮l water and half of this solution
was added instantly to the sample and the other half was
added after 2 h. The sample was then incubated O.N. at
37 ◦ C. Fifteen microliters 500 mM HCl was subsequently
added to degrade the Rapigest (incubation for 45 min at
37 ◦ C). This resulted in a turbid solution, and a pellet
could be observed after centrifugation at 14,000 rpm for
10 min. As shown by amino acid analysis, this pellet
contained amino acids in a composition resembling TeNT.
Therefore, both the supernatant and the pellet (redissolved
in 5% formic acid (FA)) were applied to chromatographic
analysis.
2.9. MALDI sample preparation
2.10. Mass spectrometers
2.8. Chromatographic analysis
The peptides resulting from the tryptic digests were
applied to RP-HPLC to separate peptides for further investigation and to compare the HPLC elution profiles. The
separations were carried out on an Äkta-Basic or Ettan HPLC
(Amersham Pharmacia Biotech, Uppsala, Sweden). The RPHPLC column was a Jupiter C18 250 mm × 2 mm, 5 ␮m,
300 Å (Phenomenex, Torrance, CA, USA). The gradient,
made of B buffer (0.05% trifluoroacetic acid (TFA) + 90%
acetonitrile (MeCN) in water), increased from 5 to 40% in
30 min; 40 to 60% in 5 min; 60 to 90% in 3 min. The A
buffer consisted of 0.06% TFA in water. The collected fractions were all analysed using matrix-assisted laser desorption
ionisation time-of-flight (MALDI TOF) MS or MALDI
quadrupole (Q) TOF MS2 .
For MALDI TOF MS analyses a Voyager-DE STR
(Applied Biosystems) or a Bruker Ultraflex (Bruker Daltonics, Bremen, Germany) with TOF-TOF technology was
used. Reflector mode was mainly activated and the all samples were analysed in positive polarity mode. Some peptides
were fragmented using the Bruker Ultraflex. The spectra were
internally calibrated when possible or, alternatively, by external calibration close to the actual target spot using a tryptic
digestion of lactoglobulin.
Modified peptides were mainly analysed by MS/MS
fragmentation using a MALDI Q-TOF Ultima (Waters/
Micromass, Manchester, UK). The collision energy during
the MS2 experiments was 70–140 eV and argon was used as
collision gas. The instrument was calibrated by multipoint
calibration using PEG 2000.
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
Peptides containing +12 Da modifications were also analysed with electrospray ionisation (ESI) Q-TOF MS and
MS2 using a Q-TOF1 hybrid instrument (Waters/Micromass,
Manchester, England) connected with a nanoESI sprayneedle.
2.11. Software
Three-dimensional structure visualization was carried out using the RasMol software version 2.7.3
(www.OpenRasMol.org) and mass spectrometric analysis
was carried out using the GPMAW software (Lighthouse
data, Odense, Denmark) [18].
2217
To investigate the parameters of the detoxification process
further, another matrix experiment was set up. Here, four concentrations of formaldehyde and lysine were chosen, giving a
4 × 4 matrix. These samples all proved non-toxic when tested
for toxicity on mice, meaning that the reaction settings had
been sufficient for a complete detoxification.
These results show that both the formaldehyde and the
lysine concentrations are critical parameters in the detoxification process and that the critical limits for obtaining detoxified
tetanus toxoid are between 1/10 and 1/4 of the standard
formaldehyde concentration (2.7–6.75 mM) and between 1/5
and 1/4 of the standard lysine concentration (2–2.5 mM).
3.2. SDS-PAGE
3. Results
3.1. Matrix experiments
Two matrix experiments were set up to study the detoxification effect of the formaldehyde and lysine concentrations.
Both parameters are known to be essential for the detoxification of TTd.
Table 1a illustrates the experimental design and lists the
toxicities of the matrix samples. Three concentrations of
formaldehyde were chosen to form the 3 × 3 matrix; the
standard concentration ‘1’ and concentrations one and two
orders of magnitude below, respectively. For each of these
concentrations, the lysine concentration was varied to three
levels; the standard concentration ‘1’ and 1/5 and 1/25 of this
(see Section 2 for actual concentrations). All samples, except
sample TTd06, were toxic, meaning that the detoxification
process failed in these samples.
The mature TeNT appears as a heavy chain of 100 kDa
disulfide bonded to a light chain of 50 kDa formed by internal
cleavage of the polypeptide chain after Ala456 [2,3]. TeNT,
TTd and TTd01–TTd24 were analysed on SDS-PAGE after
reduction of the disulfide bridges to determine the degree of
covalent cross-linking of the heavy and light chain (Fig. 2).
The non-toxic TTd01–TTd16 appeared as distinct bands in
the same region as the TTd (around 150 kDa), indicating
that the chains of these molecules had been covalently
cross-linked by at least one non-reducible covalent bond.
As expected, the TeNT chains were not cross-linked and
appeared as separated bands, which appeared below their
expected locations. Based on mass spectrometric peptide
mapping, the bands were determined to be truncated products
mainly of the heavy chain. The truncations were primarily
from the N-terminus (data not shown). This was supported
by N-terminal sequencing data on the intact toxin, which
showed multiple residues in each cycle, but yielded only a
Fig. 2. Reduced SDS-PAGE of TTd01–TTd24 including TeNT and TTd as references (right lanes). The intact, heavy and light chain mass regions are indicated.
It was observed that in particular the heavy chain of TeNT was truncated (data not shown).
2218
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
readable N-terminal sequence of the light chain (data not
shown).
The SDS-PAGE result showed that a reduction of the
formaldehyde to 1/4 of the standard conditions did not affect
the inter-chain cross-linking.
Interestingly, TTd06, TTd17 and TTd18 showed the same
degree of inter-chain linkage and moreover appeared in the
same mass region and with a similar spot-shape as the TTd
reference. These three samples were all detoxified using
the standard formaldehyde concentration. The inter-chain
linkages of TTd19–TTd21 were not complete—weak
bands appeared in the heavy and light chain mass regions.
TTd22–TTd24 were only weakly inter-chain cross-linked.
This pattern strongly suggests the formaldehyde concentration to be the most important factor for determining the
degree of cross-linking. Furthermore, the results suggest that
the lysine concentration was not critical for the inter-chain
cross-linking within the concentration range where these
experiments were carried out. Importantly, the results also
indicate that the degree of cross-linking as a single factor
was not sufficient to determine the quality of the detoxification, e.g. TTd17 and TTd18 were still toxic, despite
the fact that they appeared with the same cross-linked
bands as the non-toxic TTd06 and TTd in the SDS-PAGE
analysis. Hence, the results of this experiment show that
the cross-linking is not by itself sufficient for abolishing
toxicity.
3.3. Amino acid analysis
The composition of the TeNT, TTd and TTd01–TTd24
were analysed using amino acid analysis. With the exceptions of a few residues (lysine and tyrosine), the amino acid
compositions of all samples agreed, within the uncertainty of
the analysis, with the theoretical composition of amino acids
of TeNT and the results of a previous study [19].
The content of lysine and tyrosine showed relatively large
variations (Table 1b and c). The amount of tyrosine was
significantly reduced to 4.0–4.5% at high formaldehyde concentrations (‘2’ and ‘1’) compared to the amount of tyrosine
of the TeNT (6.0%). By lowering the formaldehyde concentration to 1/4, 1/10 and 1/100 of the standard concentration,
the amount of tyrosine increased step-wise to finally match
the levels of TeNT. This formaldehyde-induced reduction was
a result of tyrosine modification and the extent of this modification was largely independent of the lysine concentration.
Furthermore, the toxicity of the samples appeared independent of the tyrosine modification level, meaning that a high
level of tyrosine modification was not the single requirement
for a successful detoxification of the molecule (i.e. TTd17 and
TTd18 were toxic). However, there was a clear correlation
between the tyrosine modification level and the inter-chain
cross-linking observed in the SDS-PAGE analysis. In agreement with these results, it has previously been reported that
the tyrosine level is decreased in TTd (of 5.43%) compared
to TeNT (of 6.44%) [19].
The interpretation of the lysine data was complicated as
the external lysines, which are known to cross-link to the
protein, most likely were indistinguishable from the lysines
originating from the protein after acidic hydrolysis. Hence,
the exact contribution of external lysines to the total lysine
amount was difficult to determine.
Remarkably, taking the addition of external lysines to the
protein into account, the TTd and the majority of matrix
samples had a reduced amount of lysine residues compared
to the TeNT. The most likely cause is that several lysine
residues were modified by Schiff-bases to form products that
were undetectable by amino acid analysis. A small peak was
observed as a shoulder to the lysine peak corresponding to
a methylated lysine product (<5%, data not shown). In comparison, other investigators have reported that the amount of
lysine decreased to 5.51% in the TTd compared to 8.43%
in the TeNT [19]. However, their production of the TTd did
not involve lysine as an external reactant. Thus, comparing
this loss of lysines with the data from the present study gives
a measurement for the amount of external lysines that has
reacted with TTd.
There was a general correlation between the external
lysine concentration and the measured amount of lysine in the
samples (Table 1c). This correlation was most significant for
samples treated with high formaldehyde concentrations due
to the higher concentration of formaldehyde/lysine adducts
formed and became negligible for the samples treated with the
lowest formaldehyde concentrations. There seemed to be no
correlation between the amount of lysine measured and the
toxicity, e.g. the toxic TTd19 (8.39%) and TTd22 (8.69%)
showed similar amount of lysine as the non-toxic TTd06
(8.55%) and TTd (8.53%). Thus, the measured amount of
lysine cannot be used as a measure for detoxification.
3.4. Chromatographic analysis
Prior to MALDI MS analysis, peptides generated from
proteolytical digests were separated by RP-HPLC. In contrast to TeNT, which was readily digested to completion,
the digestions of the TTd and several of the matrix samples
were frustrated by molecular cross-linking, leaving a large
part of the molecule undigested. This problem was partly
circumvented using Rapigest, a surfactant, which made the
hydrophobic part of the protein more accessible. As a result
of the size of the protein, the large number of peptides generated from the proteolytic digest, resulted in chromatograms
becoming extremely complex and not all peptides were separated into individual fractions (see Fig. 3).
The various RP-HPLC elution profiles of the tryptic
digests were compared to locate differences that could reveal
cross-links and other modifications important for the detoxification process, e.g. the elution profiles of the non-toxic
TTd06 and the toxic TTd17 were compared (Fig. 3). The
elution patterns were highly similar and no significant differences could be observed. From the SDS-PAGE analysis
it was also shown that the two samples appeared as bands
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
2219
Fig. 3. Comparison of RP-HPLC elution profiles of TTd06 and TTd17. Both traces were recorded at 214 nm. SDS-PAGE results are included together with
toxicity data (inset). The chromatograms and the bands in the SDS-PAGE analysis were highly similar indicating that the toxic behaviour of TTd17 was not
caused by a modification observable in the RP-HPLC elution profile or the SDS-PAGE analysis.
with the same shape and in the same regions. Hence, the
toxic behaviour of TTd17 was not caused by a modification
observable in the RP-HPLC elution profile or the SDS-PAGE
analysis.
3.5. Tetanus toxoid modifications
The RP-HPLC separated peptides of TTd were analysed
using MALDI TOF MS and MALDI Q-TOF MS/MS. Table 2
lists the characterised TTd modifications and Fig. 4 illustrates
the observed TTd peptides and characterised modifications
mapped to the known TeNT sequence using this off-line
LC–MS approach. Eighty-seven percent of the TTd sequence
was covered and a large number of modifications were determined. Generally, partial modification was observed,
meaning that a particular TTd peptide was observed as both
the modified and unmodified variant. Most modifications
were observed in quite high ratios, usually around 1:1.
Combined with the fact that the intact toxin is truncated
this results in extremely heterogeneous TTd (Fig. 2), and
it has been impossible to obtain mass spectra of the intact
molecule.
A few +1 and +2 Da modifications appeared in the analysis
as a result of deamidations of asparagine to aspartic acids.
These deamidations were a likely consequence of sample
handling, and will not be described further here.
Two interesting mass increases were found: +12 and
+14 Da modifications, corresponding to a Schiff-base and
a methylation, respectively. Furthermore, a single +28 Da
modification was found corresponding to a dimethylation.
Unexpectedly, no cross-links were observed in the analysis.
3.5.1. Schiff-bases and degenerative products
A large number of +12 Da modifications were observed.
Fig. 5 shows three examples of fragmentations of a peptide
that was modified by +12 Da at different locations (the peptide was observed in two different HPLC fractions and called
peptide X and Y). Peptide X and Y were both N-terminally
derivatised by 4-sulfophenyl isothiocyanate (SPITC) prior to
fragmentation in order to eliminate b-ions and thereby simplifying the spectrum interpretation [21]. Peptide X (Fig. 5a),
contained a modified tyrosine, whereas peptide Y (Fig. 5b),
showed a more ambiguous fragmentation pattern with the
modification located partially on the C-terminal (most likely
the arginine residue) and partially on the internal lysine
residue. Peptide Y was also fragmented without any prior Nterminal derivatisation (Fig. 5c). This resulted, as expected, in
both the y- and b-ion series. Interestingly, as the N-terminal
was now unprotected, the +12 Da modification was exclusively located in this position. This shows that the Schiff-base
was labile and not a fixed modification. In contrast, fragmentation of peptide X without N-terminal derivatisation showed
that the +12 Da was not repositioned in this peptide (data not
shown), which indicate that the tyrosine could stabilise the
modification.
Fig. 6 illustrates the proposed structures of the +12 Da
modified residues. The structure of the tyrosine modification
is likely to resemble the previously reported structure of
the similarly modified tryptophan [22]. Here, the carbon-2
of both the indole (tryptophan) and the phenol (tyrosine)
ring links to the backbone N␣ , forming an additional
six-membered ring structure. The formation of this ring adds
sufficient stability to immobilise the modification. Alterna-
2220
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
Table 2
The modifications observed in TTd, including sequences, mass values and modification locations are listed
Mod. (Da)
Peptide
[M+H]+ unmod./mod.
Mod. location
+1
+1
+1
+2
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+12
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+14
+28
1213 DGNAFNNLDR1233
1135.6/1136.6
1381.7/1382.7
2034.0/2035.0
1135.6/1137.6
974.5/986.5
1004.6/1016.6
1264.9/1276.9
1788.9/1800.9
2411.1/2423.1
2002.1/2014.1
2062.9/2074.9
1405.9/1417.9
1476.8/1488.8
1703.9/1715.9
778.5/790.5
850.5/862.5
1194.7/1206.7
1520.8/1532.8
2566.2/2578.2
2721.4/2733.4
3498.6/3510.6
1852.9/1864.9
644.4/656.4
863.5/875.5
1052.7/1064.7
1232.7/1244.7
1378.8/1390.8
1397.8/1409.8
676.5/690.5
1075.6/1089.6
1135.1/1149.1
1169.5/1183.5
1291.7/1305.7
1293.7/1307.7
1312.6/1326.6
1336.7/1350.7
1443.7/1457.7
1563.9/1577.9
1766.1/1780.1
1976.1/1990.1
2394.2/2408.2
2557.3/2571.3
2580.3/2594.5
2668.3/2682.3
2857.4/2871.6
3007.5/3021.5
1419.8/1447.8
N
N
N
N
Nterm
Nterm
Nterm
Nterm
Nterm
Nterm , Y, K, R
Y, K
Y
Y
Y
K
K
K
K
K
K
K
K, R
R
R
R
R
R
R
K
K
K
K
K
K
K
K
K
K
K
K
K
K
K
K
K
K
K
335 DSNGQYIVNEDK345
1261 NASLGLVGTHNGQIGNDPNR1280
1213 DGNAFNNLDR1222
1 PITINNFR8
1115 DFWGNPLR1122
1030 WVFITITNDR1039
921 AIHLVNNESSEVIVH936
384 IPNLLDDTIYNDTEGFNIESK404
511 IIVDYNLQSKITLPNDR527
329 YQFDKDSNGQYIVNEDK345
564 LYAQKSPTTLQR574
1103 LYTSYLSITFLR1114
754 IIDYEYKIYSGPDK767
366 FNIKTR371
35 AFKITDR41
1246 DLKTYSVQLK1255
723 AKWLGTVNTQFQK735
300 AIANKLSQVTSCNDPNIDIDSYK322
1115 DFWGNPLRYDTEYYLIPVASSK1136
49 YEFGTKPEDFNPPSSLI. . .YDPNYLR78
1073 LDRCNNNQYVSIDKFR1089
1244 LRDLK1248
1167 RLYNGLK1173
441 IIPPTNIRE449
1012 QITFRDLPDK1021
1005 DSAGEVRQITFR1016
38 ITDRIWIVPER48
1174 FIIKR1178
1002 TLKDSAGEVR1011
702 TIDNFLEKR710
1295 DKILGCDWY1303
1226 VGYNAPGIPLYK1237
743 SLEYQVDAIKK753
177 VDNKNYFPCR186
814 KQLLEFDTQSK824
248 SHEIIPSKQEIY259
281 LISIDIKNDLYEK293
156 LIIFGPGPVLNKNEVR171
1099 EIEKLYTSYLSITFLR1114
346 FQILYNSIMYGFTEIELGKK365
1191 SGDFIKLYVSYNNNEHIVGYPK1212
300 AIANKLSQVTSCNDPINDIDSYK322
1256 LYDDKNASLGLVGTHN. . .DPNR1280
305 LSQVTSCNDPINDIDSYKQIIQQK328
575 ITMTNSVDDALINSTKIYSYFPSVISK601
1226 VGYNAPGIPLYKK1238
Partial modifications were observed for almost all the reported locations.
tively, the +12 Da modification of tyrosine is a residual mass
of degenerative processes of lysine/formaldehyde adducts or
cross-links during MS. The mass increase of arginine may
also be ascribed to this degenerative fragmentation, as the
guanidinium group does not behave as a common amino
group. Thus, it is not expected that arginine can form Schiffbases. However, the observation of a +12 Da modification on
arginine in Fig. 5b indicates the opposite. The labile nature
of the Schiff-bases on lysine residues can be ascribed to the
lack of stabilizing ring structures. The previously reported
ring formation of the +12 Da modified N-terminus [9]
introduces some stability into the peptide and explains this
preferential location in Fig. 5c. However, the ring structure
also posses some lability as the N-termini of the peptides
can be readily modified, e.g. by SPITC (Fig. 5b). N-terminal
Edman degradation of TTd (data not shown) provided no
sequence information, indicating that all N-termini were
modified and that the ring formation was either stabilised
on the intact protein or by Edman degradation. These data,
however, showed a substantial amount of lysine in the
first few cycles indicating that cross-linked external lysine
residues were released during the Edman cycle.
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
2221
Fig. 4. Sequence coverage of the observed TTd peptides mapped to the known TeNT sequence. Residues in bold represent unmodified residues, whereas
residues marked with circles and squares represent Schiff-base modifications and methylations, respectively. Residues 1–457 constitute the TeNT light chain
and residues 458–1315 constitute the TeNT heavy chain. The location dividing the two chains is indicated. Note that several modifications (visualised on
isoleucine, alanine and aspartic acid) are located on the N-termini of the resulting peptides after trypsin digestion. The presented modifications are based on
the complete set of observed TTd modifications (Table 2).
It can be speculated that a fraction of the observed +12 Da
modified peptides were MS-mediated fragments of crosslinked products. This is supported by MS data from a gel
filtration of tryptic digested TTd (Fig. 7a). Here, several small
+12 Da modified peptides were observed together with their
corresponding unmodified peptides in early eluting fractions.
Normally, the gel filtration does not allow such small peptides to elute in these fractions, suggesting that these peptides
were not naturally occurring in the sample, but rather crosslinked peptides that were separated when analysed in MS.
Hence, it is proposed that the stability of the methylene bridge
connecting the two primary amines of the two peptides was
insufficient to survive the MALDI MS analysis and consequently fell apart in the instrument. Following this line of
thought, the +12 Da linker could be located on either peptide,
giving rise to the unmodified and the +12 Da modified variant of both peptides involved in the cross-link (Fig. 7b). This
fragmentation might explain the lack of characterised crosslinks in the MS analysis. Being a milder ionisation technique,
ESI MS was also performed on the +12 Da modified peptides
to observe non-fragmented cross-links. However, this yielded
the same results as the MALDI analysis (data not shown).
In addition, two synthetic peptides (Ac-LNLLYLLNLCONH2 and Ac-LNLLRLLNL-CONH2 ) were treated with
formaldehyde and lysine according to ref. [9]. MALDI and
ESI MS were performed in order to establish whether the
+12 Da modifications on tyrosine and arginine were residual
mass increases of lysine/formaldehyde adducts, cross-links
degenerated during MS or simple +12 Da modifications. Neither of the two peptides was modified as a result of this
treatment (data not shown). This indicates that the initial reaction with formaldehyde has to take place on an amine, and the
+12 Da modifications of arginine and tyrosine are the results
of a secondary reaction or cross-link. Additionally, the conformation of the protein or surrounding amino acid residues
may play a role in the modification of these residues.
3.5.2. Methylations
Numerous methylations were characterised in the TTd
molecule (Table 2). These were all located on lysine residues.
With the exception of a single dimethylation, the lysine
residues were all monomethylated and the majority of these
were located internally in the tryptic peptides, indicating that
trypsin did not recognise this residue when modified.
The methylations were most likely the result of reduction
of the formaldehyde-induced lysine Schiff-bases (Fig. 1) and
the dimethylated lysine was generated by an additional cycle
of Schiff-base formation and reductive methylation.
4. Discussion
In recent years, several studies have been performed on
formaldehyde-treated diphtheria and tetanus toxoids, e.g.
refs. [15,23–25]. These investigations have mainly been dedicated to establish parameters, which could be used for
2222
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
2223
Fig. 6. The proposed structure of the modified tyrosine, based on the observed stability of this residue, is shown together with the known structure of the
similarly modified tryptophan [22]. The methylene bridge links carbon-2 and backbone N␣ in both structures. In addition, the labile Schiff-bases are shown
together with the imidazolidinone adduct of the N-terminus [9]. Alternatively, the observed +12 Da mass increases of tyrosine and arginine residues can be the
residual masses of lysine/formaldehyde adducts or cross-links, which are degenerated during MS.
Fig. 7. (a) (Inset) Gel filtration of tryptic digested TTd. The fraction analysed by MS is indicated. MALDI TOF MS of the fraction collected from the gel
filtration. Surprisingly, several peptide pairs (unmodified and +12 Da variants) were observed in this fraction. No large peptides were observed in the high
mass region (region not shown). (b) Proposed MS fragmentation of methylene bridged cross-linked peptides. The fragmentation can occur either at the ␣- or
␤-location giving rise to both the unmodified and the +12 Da modified variant of each peptide. P1 and P2 represent different peptide units.
monitoring the quality of the toxoid. This is important for
both ethical and economical reasons because the development of a set of in vitro methods for predicting the potency
and quality of a newly produced toxoid can lead to reduction
in the number of animals used for these tests. However, no
functional in vitro test has yet been accepted as an alternative for the potency determination of these toxoids, as it is
extremely complicated to imitate the immune response.
Fig. 5. MALDI Q-TOF MS/MS fragmentation of three variations of the same peptide eluting from the RP-HPLC in two different positions. (a) Peptide X was
SPITC derivatised and fragmented. The y-ions clearly showed the sequence and revealed that the +12 Da modification was located on the tyrosine residue. (b)
Peptide Y was also derivatised with SPITC and fragmented. y-ion pairs (e.g. m/z 501/513, 614/626) were observed in the low mass region, indicating that one
of the C-terminal residues (most likely the arginine residue) was partially modified with a Schiff-base. y-ions containing the internal lysine (y8 , y9 and higher)
were completely modified, indicating that this lysine residue also was partly modified. (c) Peptide Y was fragmented directly without prior derivatisation. Nearly
complete y- and b-ion series were observed when the N-terminus was modified with +12 Da to create theoretical fragments that matched the experimental values.
2224
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
By varying the active components (formaldehyde and
lysine) during detoxification we tried to determine the limits for a non-toxic compound. In order to establish the
mechanism for detoxification we performed protein chemical analysis on all samples including the intact toxin and
commercial toxoid.
4.1. Matrix experiments
By combining the results of the toxicity tests, SDSPAGE, amino acid analysis, RP-HPLC and MS performed on
TTd01–TTd24 and using commercial TTd and native TeNT
as references, a substantial amount of important information
could be obtained.
The SDS-PAGE results showed that inter-chain crosslinking of the TTd was dependent on the formaldehyde
concentration. This was expected, as formaldehyde is known
to add to the ␧-amino group of lysine residues forming Schiffbases, and thereby being a key element in cross-linking.
Remarkably, combining these SDS-PAGE analyses with the
toxicity data, it could be concluded, that the inter-chain
cross-linking was not the single factor determining the detoxification of the molecule. As cross-linking of TTd should
prevent the transfer of the light chain to the cytosol where
its substrate (synaptobrevin) is located, it was somewhat surprising that even extensive cross-linking was not sufficient
for abolishing toxicity.
The toxicity data showed that a successful detoxification could be maintained when the formaldehyde and lysine
concentrations were reduced to 1/4 of the standard levels,
indicating that the standard production of TTd was carried
out well above the critical limit. Further reduction of the concentrations illustrated that both formaldehyde and lysine were
critical parameters of the detoxification.
The tyrosine quantity of the matrix samples provided some
useful information. Even though the toxicity appeared independent of the tyrosine modification level, there was a clear
correlation between the amount of tyrosine and the degree
of inter-chain cross-linking of the molecules. The amount of
lysine in the matrix samples was reduced compared to the
TeNT. This was unexpected, as external lysines are known
to cross-link to the protein in the presence of formaldehyde
through lysine–lysine bonds.
It was expected that comparison of the HPLC elution
profiles of the matrix samples would reveal some distinct differences that could pinpoint cross-links and other significant
modifications responsible for the detoxification. In particular,
it was interesting to compare the elution profiles of TTd06 and
TTd17/TTd18 as both samples appeared with the same degree
of inter-chain cross-linking in gels, but expressed marked differences in toxicity. Hence, any difference between these two
samples was potentially the modification defining the detoxification of TTd06 and leaving TTd17/TTd18 toxic. However,
no differences could be observed between the two elution profiles (Fig. 3), indicating that the modifications responsible for
the detoxification either did not alter the retention time of the
peptide, broke down under the conditions of RP-HPLC or
were present in undetectable low amounts.
4.2. Mapping onto the three-dimensional structure
A few larger peptides were not observed in the off-line
LC–MS analysis in spite of the peptides being amenable for
mass spectrometric analysis (see Fig. 4). The small peptides
missing in the peptide map are likely lost due to elution in the
solvent front of the HPLC. The reason for not observing these
peptides cannot be due to incomplete digestion, as the neighbouring peptides were readily observed, and only two of the
peptides (residues 111–127 and 655–688) are significantly
more hydrophobic than the average peptide.
Mapping the larger peptides and the identified modifications onto the known 3D structures of the TeNT light chain
and the C-terminal part of the heavy chain revealed some
interesting features (Fig. 8).
The two peptides missing in the light chain (residues
111–127 and 261–281) are both alpha-helical regions that
transverse the entire globular domain of the light chain, but
their main surface exposure are on the same face of the
molecule close to the active site. The peptide missing in
the C-terminal part of the heavy chain is a three-␤ strand
surface region. The location of the remaining three missing
peptides in the heavy chain is unknown as no structure is
known for this part. A common feature of the three structuremapped peptides is the presence of a surface-located tyrosine
residue (Tyr122, Tyr265 and Tyr909). As tyrosine is known
to generate stable formaldehyde-induced cross-links [6] an
explanation for the missing peptides could be that they are
cross-linked and either present in low amounts or too heterogeneous to be detected. Of the remaining missing peptides,
the N-terminus of the heavy chain could be missing due to
trimming of the chain (see Fig. 2) and the remainder all have
a tyrosine residue located among hydrophilic residues, thus
expected to be surface located.
We therefore suggest that the light and heavy chain of
TeNT interact or are in close contact in two locations, close
to Tyr122 and Tyr265 in the light chain, and close to Tyr122,
Tyr265 and/or Tyr909 in the heavy chain.
Although most lysine residues are freely exposed on the
surface of the molecule, most of the observed methylated
lysines are observed in two regions: in the heavy chain, four
out of five methylated residues are located close together
while in the light chain, six out of seven methylations are
located next to the ‘missing’ peptides, four on the face
hypothesized to be interfacing the heavy chain.
4.3. Biological relevance of observed modifications
Some of the lysine residues were incompletely modified
and observed partly as unmodified residues and partly as
Schiff-bases or methylated residues. In a study performed
several decades ago, it was reported that lysine monoand dimethylation of TeNT could reduce toxicity [26].
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
2225
Fig. 8. Space-fill model of TeNT light and heavy chain (C-terminal part). (A) The light chain is shown with a view of the active site. The large unmapped
peptides are highlighted along with the observed modifications of methylated lysine residues and Schiff-bases. The C-terminal part of the heavy chain is shown
with the face containing the large missing peptide. The C-terminus of the light chain and the N-terminus of the heavy chain are indicated. Cys438 in the light
chain is known to disulfide bond to Cys466 in the heavy chain—neither is part of the known structures. (B) The two fragments in A rotated 180◦ . Note that the
mapped modifications are based on the complete set of observed TTd modifications (Table 2), and that these are partial modifications.
Albeit greatly reduced, the toxicity was never completely
eliminated. In contrast to carbamylation, which was reported
to destroy the toxicity completely, the remaining toxicity
of the reductive methylation was related to the fact that the
substitution of a methyl group for hydrogen in the ␧-amino
group did not introduce significant charge and stereochemical alterations of the lysine side-chain. The spacefill model
(Fig. 8), shows that a methylated lysine is located at the
entrance to the active site. Here, a small change in the
charge environment or additional steric hindrance produced
from modified side-chains might be sufficient to limit the
substrate accessibility and thereby reduce the toxicity.
In another study, the effect of tyrosine modification on
TeNT toxicity was investigated and it was reported that when
only a few nitrations of the tyrosine residues were introduced,
the tetanus toxin was rendered non-toxic and immunogenic
[27]. In relation, a few modified tyrosine residues were
observed in this study, generating the stable compound shown
in Fig. 6. Even though this modification did not introduce
a charge alteration (unlike the nitration), the modification
2226
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
might still affect the generation of stable tyrosine-lysine
cross-links. Whether this is relevant for the detoxification
of TTd is unknown, however, the results from this and other
studies indicate that the tyrosine residues take a part in the
expression of the toxicity either directly or indirectly.
The active site of the TTd light chain (His232–His236)
and the primary structure surrounding it was characterised as
an unmodified region. Thus, it can with some assumptions
be ruled out that it was the modification of the active site
that destroyed the toxicity of the molecule. These assumptions included that Zn2+ , known as an essential metal ion for
TeNT activity [28], was present in the TTd and that the toxin
and toxoid had similar conformations. The latter is shown to
be true for diphtheria, where only limited spatial changes are
observed in the toxoid [24]. Hence, it is likely not the alteration of the active site that is essential for the detoxification
of the TTd.
In addition to the methylations described above, a number
of Schiff-bases were also located to peptides/residues at the
surface of the structure (Fig. 8). In theory, the Schiff-bases
are known to react readily with several amino acid residues.
This reactivity enable them to form cross-links to a large
number of compounds, either in the vaccine or when injected
into the individual. It can be speculated that cross-linking to
other proteins might eliminate one or more of the three TeNT
functions: (i) binding and internalization into the neuron (Cterminal of heavy chain), (ii) translocation of the light chain
across the vesicle membrane into the cytosol (N-terminal of
heavy chain) and (iii) cleavage of synaptobrevin (light chain).
Destroying one of these functions should completely inactivate the TeNT. In relation to this, it can also be predicted that
the toxicity is partly lost due to the inter-chain cross-linking
of the TTd (SDS-PAGE; Fig. 2) as this could prevent the
translocation of the light chain to the cytosol.
Recently, Jiskoot and co-workers reported numerous
formaldehyde-induced modifications when analysing reactions of formaldehyde with model peptides and a small
protein [9,10]. Here, they found several peptide–glycine
products and other cross-links using ESI MS. Upon fragmentation these peptides showed +12 Da fragmentation products.
However, when we performed ESI MS on the +12 Da modified peptides, no cross-links were observed, i.e. results
were identical to MALDI MS analysis. When comparing
results from these related studies, it should be taken into
account that the conditions under which these studies were
performed were slightly different, e.g. 80 mM glycine and
formaldehyde was used compared to the use of 10 mM lysine
and 27 mM formaldehyde in this present study. The higher
concentrations are likely change the nature of the chemistry
processes and induce more modifications. However, this does
not entirely explain the very limited number of modifications
observed in the TTd. It cannot be excluded that a greater
number of modifications were present in the TTd but not
detected due to the off-line LC–MS approach. Every significant signal not matching a theoretical value was fragmented
using MS/MS, but the very complex LC fractions increased
the chance of missing important modified components due
to suppression in MS or heterogeneity of the modifications.
In particular we looked for formaldehyde/lysine additions,
because the amino acid analyses showed the addition of this
compound was crucial for detoxification. The mass increase
of a single formaldehyde/lysine adduct to most residues in
peptides is 158 Da (depending on which residue is modified).
However, this adduct might be further modified, and hence
it is very difficult to predict modifications by mass values
alone.
The results indicate that the nature of the formaldehydeinduced modification reaction is likely to depend on the
conformation of the protein, as both the accessibility of reactants (formaldehyde and lysine) and the spatial location of
the amino acid residues are parameters of importance.
Based on results presented here and in other studies, the
role of formaldehyde as a cross-linking agent in the detoxification process is well established. However, the function of
lysine residues (or glycine) in the reaction matrix is more
ambiguous. A likely effect of the amino acid residues is
that they stabilise the observed Schiff-bases, which then contribute to the detoxification of TTd.
In summary, two scenarios will result in loss of TeNT
toxicity. First, the TTd can be modified in a way that it
never reaches the location of its substrate (synaptobrevin).
Secondly, the modification can prevent it from cleaving
the substrate. The scenario responsible for the toxicity loss
remains so far unknown. It is estimated that both the Schiffbases and the lysine methylations are capable of significantly
reducing the tetanus toxin activity. However, it is unknown
whether these modifications are sufficient for a complete
elimination of toxicity. In addition, it is likely that the interchain cross-linking is involved directly or indirectly in the
detoxification. Recently, several studies have focused on
developing synaptobrevin-based assays both for the tetanus
neurotoxin quantification and for the in vitro determination
of specific toxicity in tetanus vaccines [29–31]. Here, it has
been reported that active components are still present in the
vaccine samples, which indicate that at least part of the toxicity loss is linked to the alteration of toxin to a compound
that never reaches it substrate.
Altogether, further studies have to be carried out to elucidate the complete detoxification mechanism. These results
will, together with the information provided from this investigation, represent a fundamental basis of knowledge for
designing new methods and strategies for determining the
quality of TTd and other toxoids by means of in vitro techniques.
Acknowledgements
We thank Dr. Thomas J.D. Jørgensen for fruitful discussions on chemical related topics. Dr. Paul Robert Hansen
is gratefully acknowledged for synthesis of two peptides
for the analysis of formaldehyde chemistry. This work was
M. Thaysen-Andersen et al. / Vaccine 25 (2007) 2213–2227
supported by a grant to M. Thaysen-Andersen from the
Oticon Foundation (Hellerup, Denmark).
References
[1] Popoff MR. Ecology of neurotoxigenic strains of Clostridia. Curr Top
Microbiol Immunol 1995;195:1–29.
[2] Eisel U, Jarausch W, Goretzki K, Henschen A, Engels J, Weller U, et al.
Tetanus toxin: primary structure, expression in E. coli, and homology
with botulinum toxins. EMBO J 1986;5:2495–502.
[3] Fairweather NF, Lyness VA. The complete nucleotide sequence of
tetanus toxin. Nucleic Acids Res 1986;14:7809–12.
[4] Ramon G. Sur la toxine et sur l’ anatoxine diphtériques. Ann Inst Pasteur
1924;38:1–10.
[5] Fraenkel-Conrat H, Olcott HS. Reaction of formaldehyde with proteins.
V. Cross-linking amino and primary amide or guanidyl groups. J Am
Chem Soc 1948;70:2673–84.
[6] Fraenkel-Conrat H, Olcott HS. Reaction of formaldehyde with proteins.
VI. Cross-linking of amino groups with phenol, imidazole or indole
groups. J Biol Chem 1948;174:827–43.
[7] Blass J, Bizzini B, Raynaud M. Mechanism of detoxification by formol.
C R Acad Sci Hebd Seances Acad Sci D 1965;261:1448–9.
[8] Blass J. Preparation et etude dune base de mannich cristallisee obtenue
par action du formol sur un melange de threonine et de dimethyl-2,4phenol. Bull Soc Chim Fr 1966;10:3120–1.
[9] Metz B, Kersten GF, Hoogerhout P, Brugghe HF, Timmermans HA,
de Jong A, et al. Identification of formaldehyde-induced modifications in proteins: reactions with model peptides. J Biol Chem
2004;279:6235–43.
[10] Metz B, Kersten GFA, Baart GJ, Jong AD, Meiring H, Hove JT, et
al. Identification of formaldehyde-induced modifications in proteins:
reactions with insulin. Bioconjug Chem 2006;17:815–22.
[11] Kreeftenberg JG. Developments in the reduction refinement and
replacement of aminal tests in the quality control of immunobiologicals.
Dev Biol Stand 1999;101:13–7.
[12] World Health Organization Expert Committee on Biological Standardization. Requirements for diphtheria, tetanus, pertussis and combined
vaccines. Geneva: WHO; 1990.
[13] Castle P. Combined vaccines: policy and practice in the European Pharmacopoeia. Biologicals 1994;22:381–7.
[14] Gupta RK, Siber GR. Reappraisal of existing methods for potency testing of vaccines against tetanus and diphtheria. Vaccine 1995;13:965–6.
[15] Metz B, Hendriksen C, Jiskoot W, Kersten G. Reduction of aminal use
in human vaccine quality control: opportunities and problems. Vaccine
2002;20:2411–30.
[16] European Pharmacopoeia—Supplement 2001. Assay of hepatitis B
vaccine (rDNA). Strasbourg: Counsil of Europe; 2001. p. 104.
2227
[17] Strupat K, Karas M, Hillenkamp F. 2,5-Dihydroxybenzoic acid: a new
matrix for laser desorption—ionization mass spectrometry. Int J Mass
Spectrom Ion Processes 1991;111:89–102.
[18] Peri S, Steen H, Pandey A. GPMAW—a software tool for analysing
proteins and peptides. Trends Biochem Sci 2001;26:687–9.
[19] Bizzini B, Blass J, Turpin A, Raynaud M. Chemical characterization
of tetanus toxin and toxoid. Amino acid composition, number of SH
and S–S groups and N-terminal amino acid. Eur J Biochem 1970;17:
100–5.
[21] Keough T, Youngquist RS, Lacey MP. A method for high-sensitivity
peptide sequencing using postsource decay matrix-assisted laser desorption ionization mass spectrometry. Proc Natl Acad Sci USA
1999;96:7131–6.
[22] Cain M, Weber RW, Guzman F, Cook JM, Barker SA, Rice KC,
et al. Beta-carbolines: synthesis and neurochemical and pharmacological actions on brain benzodiazepine receptors. J Med Chem
1982;25:1081–91.
[23] Metz B, Jiskoot W, Hennink WE, Crommelin DJ, Kersten GF. Physicochemical and immunochemical techniques predict the quality of
diphtheria toxoid vaccines. Vaccine 2003;22:156–67.
[24] Paliwal R, London E. Comparison of the conformation, hydrophobicity, and model membrane interactions of diphtheria toxin to those
of formaldehyde stabilization of the native conformation inhibits
changes that allow membrane insertion. Biochemistry 1996;35:2374–
9.
[25] Kersten G, Jiskoot W, Hazendonk T, Spiekstra A, Westdijk J, Beuvery
C. Characterization of diphtheria toxoid. Pharm Pharmacol Commun
1999;5:27–31.
[26] Robinson JP, Picklesimer JB, Puett D. Tetanus toxin. The effect of
chemical modifications on toxicity, immunogenecity and conformation.
J Biol Chem 1975;250:7435–42.
[27] Bizzini B, Turpin A, Raynard M. Immunochemistry of tetanus toxin.
The nitration of tyrosyl residues in tetanus toxin. Eur J Biochem
1973;39:171–81.
[28] Schiavo G, Poulain B, Rossetto O, Benfenati F, Tauc L, Montecucco
C. Tetanus toxin is a zinc protein and its inhibition of neurotransmitter release and protease activity depend on zinc. EMBO J
1992;11:3577–83.
[29] Perpetuo EA, Luliano L, Prado SM, Fratelli F, Fernandes I, Lebrun I.
Development of an operational synaptobrevin-based fluorescent substrate for tetanus neurotoxin quantification. Biotechnol Appl Biochem
2002;36:155–61.
[30] Kegel B, Bonifas U, Silberbach K, Kramer B, Weisser K. In vitro
determination of specific toxicity in tetanus vaccines. Dev Biol
2002;111:27–33.
[31] Kegel B, Bonifas U, Behrensdorf-Nicol H, Klimek J, Silberbach K,
Weisser K, et al. Development of an in vitro assay for testing the safety
of tetanus vaccines. In: Poster at the fifth world congress on alternatives
and aminal use in the life sciences. 2005.