A Whole Proteome Inventory of Background Photocrosslinker Binding

Article abstract of DOI:10.1002/anie.201605993

Affinity-based protein profiling (AfBPP) is a widely applied method for the target identification of bioactive molecules. Probes containing photocrosslinkers, such as benzophenones, diazirines, and aryl azides, irreversibly link the molecule of interest t

Full text of DOI:10.1002/anie.201605993

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201605993
German Edition: DOI: 10.1002/ange.201605993
Photoaffinity Labeling
AWhole Proteome Inventory of Background Photocrosslinker Binding
Philipp Kleiner+, Wolfgang Heydenreuter+, Matthias Stahl+, Vadim S. Korotkov, and
Stephan A. Sieber*
Abstract: Affinity-based protein profiling (AfBPP) is a widely
applied method for the target identification of bioactive
molecules. Probes containing photocrosslinkers, such as ben-
zophenones, diazirines, and aryl azides, irreversibly link the
molecule of interest to its target protein upon irradiation with
UV light. Despite their prevalent application, little is known
about photocrosslinker-specific off-targets, affecting the reli-
ability of results. Herein, we investigated background protein
labeling by gel-free quantitative proteomics. Characteristic off-
targets were identified for each photoreactive group and
compiled in a comprehensive inventory. In a proof-of-principle
study, H8, a protein kinase A inhibitor, was equipped with
a diazirine moiety. Application of this photoprobe revealed, by
alignment with the diazirine background, unprecedented
insight into its in situ proteome targets. Taken together, our
findings guide the identification of biologically relevant bind-
ers in photoprobe experiments.
recent study systematically identified off-targets of the three
photoreactive groups described above embedded in a minimal
scaffold on 2D-SDS gels.[7] Previous work in our laboratory
has also indicated a number of ligand-independent photo-
crosslinker-specific protein hits.[8] However, whereas some of
them correlated well with the reported gel-based off-targets,
others seemed to be detected solely by quantitative gel-free
procedures—the state-of-the-art technique in chemical pro-
teomics.
To obtain a comprehensive picture focused on the needs
of quantitative and gel-free target identification, we herein
report a full inventory of benzophenone, aryl azide, and
diazirine off-targets (photomes) in human cell lines by
quantitative high-resolution proteomics. Guiding principles
to understand off-target preferences and minimize non-
specific labeling are also provided. Finally, we demonstrate
the value added by our photome profiling study with
a photoprobe based on H8, a protein kinase A (PKA)
inhibitor. The conclusions drawn from this experiment
would have been heavily biased by photome hits without
our off-target exclusion list.
Our strategy for compiling an inventory of the non-
specific background proteome began with the chemical design
and synthesis of basic photocrosslinker probes. All three
photoreactive groups, aryl azide (AA), benzophenone (BP),
and diazirine (DA), were embedded in a simple molecular
scaffold to minimize ligand-specific protein interactions. In
the case of AA and BP, we appended the alkyne tag, which is
required for target identification by click chemistry (CC;
Figure 1B),[9] to an aryl ring with a small aliphatic ether
(AA-1) or an amide linkage (BP-1), respectively (Figure 1A).
By definition, AA- and BP-containing photoprobes have an
aryl ring in their minimal molecular scaffolds, and for
comparison, we designed a minimal aromatic DA probe
(DA-1) in a similar way. To account for photocrosslinker-
independent background labeling, we selected the DA-
1 scaffold lacking the diazirine group (NC-1; Figure 1C).
All molecules were prepared according to established proce-
dures with slight modifications (see the Supporting Informa-
tion, Schemes S1 and S2).
To evaluate the proteome binding of AA-1, BP-1, and
DA-1, we chose two human cell lines, A549 and HeLa, as
representative and commonly used test systems. The photo-
crosslinker probes were incubated with intact cells for 1 h and
subsequently irradiated for 30 min with UV light; these
conditions are similar to most common AfBPP labeling
procedures. In addition, one sample was treated with the
NC-1 control under identical conditions. First, we investigated
the cell permeability of all molecules by gel-based analysis of
their labeling pattern in the cytosolic and membrane fractions.
Interestingly, all photoprobes predominantly labeled cyto-
P
hotocrosslinkers are important tools for chemical biology
research and mediate irreversible linkages between proteins
and ligands upon UV irradiation, for example.[1] A prominent
application is affinity-based protein profiling (AfBPP) where
small molecules are equipped with a photocrosslinker to
facilitate irreversible binding to
a
protein target.[1b,2]
Researchers have the choice between three major photo-
reactive groups, which differ in their size, efficacy, and
possible side reactions, namely benzophenones, aryl azides,
and diazirines (Figure 1A).[1c] Alkyl diazirines,[3] in particular,
have recently experienced a renaissance in application[4]
owing to their small size and high crosslinking efficiency.
Furthermore, several studies comparing the reactivity and
efficiency of these three photocrosslinkers reported superior
properties for diazirines, further encouraging their use.[5]
Despite this popularity, very little is known about photo-
crosslinker-associated off-target binding. Thus control reac-
tions such as competition with the parent, unmodified ligand
are recommended;[6] however, these experiments are chal-
lenging owing to the irreversible nature of the photoprobe–
protein interaction while the competing parent compound
binds reversibly. In addition, a minimal photoprobe lacking
the ligand-specific structure can be applied as a control.[4b]
A
[*] M. Sc. P. Kleiner,[+] Dr. W. Heydenreuter,[+] M. Sc. M. Stahl,[+]
Dr. V. S. Korotkov, Prof. Dr. S. A. Sieber
Center for Integrated Protein Science at the Department of Chemistry
Technische Universitꢀt Mꢁnchen
Lichtenbergstrasse 4, 85747 Garching (Germany)
E-mail: stephan.sieber@tum.de
[+] These authors contributed equally to this work.
Supporting information for this article can be found under:
http://dx.doi.org/10.1002/anie.201605993.
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 1. Analysis of photocrosslinker background binding. A) Photocrosslinker probes used in this study. The photoreactive group is highlighted
in blue. B) Gel-free photolabeling strategy using SILAC to quantify probe hits over DMSO control. See the Methods Section in the Supporting
Information for details. C) Structure of the negative control probe (NC-1) used for the identification of non-photospecific target proteins.
D) Volcano plots of enriched proteins in SILAC experiments with the photocrosslinkers AA-1, BP-1, and DA-1 and the control NC-1. A log2
enrichment of >1 implies a double excess of protein in the probe sample compared to the DMSO control. A p-value of <0.05 was required to
classify a hit as significantly enriched (highlighted in red). E) Venn representation showing proteins that were significantly enriched in all four
experiments.
solic proteins, suggesting sufficient cell permeability (Fig-
ure S1). BP-1 stood out as the probe with the most intense
labeling and the highest number of targets while AA-1 and
DA-1 exhibited much reduced proteome labeling.
To obtain a more comprehensive inventory of background
binders, we switched to gel-free, quantitative proteomics. This
technique has become the gold standard in AfBPP, and its
high sensitivity requires rigorous characterization of non-
specific binders. We thus applied the stable isotope labeling of
amino acids in cell culture (SILAC) approach and treated
cells labeled with heavy or light isotopes with probe (3 mm) or
DMSO, respectively.[10] Cells were lysed, equal protein
amounts from DMSO and probe-treated samples pooled,
and whole proteomes clicked to biotin azide. After enrich-
ment and tryptic digest, the samples were analyzed by high-
resolution liquid chromatography tandem mass spectrometry
(LC-MS/MS; Figure 1B); log2(isotope ratios) of 0 indicate no
enrichment compared to the DMSO control, and log2 ratios of
> 1 with a p-value of < 0.05 were regarded as significant hits
(Figure 1D and Figure S2). Whereas NC-1 showed only
marginal protein binding, all three photoprobes exhibited
significant protein enrichment that was largely probe-specific.
For example, BP-1, previously regarded as a selective photo-
linker owing to its reversible mode of photoactivation,[11]
exhibited the highest number of protein targets (114) in
A549 cells followed by DA-1 (19) and AA-1 (5; Figure 1E).
2
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
Interestingly, background labeling in HeLa cells was more
most of these proteins were not identified by gel-based off-
evenly distributed with 46, 78, and 85 hits for AA-1, BP-1, and
DA-1, highlighting the importance of cell-line-specific off-
target analysis (Figure S2). A list of all identified proteins is
provided in the Supporting Information.
target profiling,[7] and furthermore, hits of gel studies are
underrepresented here, confirming the relevance of the
proteomic preparation technique for the results of AfBPP
studies.
Considering the increasing application of DA photo-
probes in chemical biology, we synthesized three additional
minimal probes for in-depth analysis (Figure 1A and
Scheme S3). Here, the DA moiety was embedded in different
aliphatic or aromatic scaffolds with the aim of dissecting off-
target binding and connecting protein hits to structural motifs.
All four DAs (DA-1–DA-4), including a close mimic of
a previously reported control compound,[4b] were applied in
gel-based cellular protein profiling. All probes labeled intra-
cellular targets, suggesting sufficient cell permeability (Fig-
ure S3).
Subsequent gel-free, quantitative proteomics studies
revealed a significant influence of the scaffold on protein
binding (Figure S4). Probe DA-2, which contains an aryl
piperazine, exhibited the highest labeling in A549 cells
followed by the aromatic probe DA-1 (Figure 2A). In HeLa
cells, DA-1 and aliphatic DA-4 contributed to the majority of
hits (Figure 2B). Interestingly, DA-3, with an amide bond and
a short aliphatic chain, was most selective in both cell lines,
suggesting that affinity probe design could benefit from such
short alkyl linkers.
To gather comprehensive information on the most prom-
inent DA-specific background targets, we overlaid the
volcano plots for all probes in A549 and HeLa cells. DA-
specific protein targets should thus be extensively enriched
whereas scaffold-specific hits relocate around zero (Fig-
ure 2C; for a detailed explanation, see the Supporting
Information). This mathematical operation revealed a set of
eight and seven significant proteins for A549 and HeLa cells,
respectively, after Benjamini–Hochberg correction (Fig-
ure 2D,E). Satisfyingly, 291 out of 1655 protein groups in
the A549 proteome were annotated as membrane-bound.
Identified hits that are predominantly localized in the
lysosome and mitochondrion can be grouped into three
superfamilies: 1) channels or channel-associated proteins
(e.g., VDAC1 and VDAC2), 2) catabolic enzymes (e.g.,
CTSD), and 3) small-molecule binders (e.g., ALDH1B1).
For example, VDAC1 and VDAC2 are highly abundant
channels with permeability for small molecules. Similarly,
ECH1 and ALDH1B1 are catabolic enzymes known to bind
small molecules. A summary of all non-specific photoprobe
targets is provided in Figure 2F and Table S1. To account for
label-free MS quantification, a rising proteomic technique
independent of SILAC, we applied all DA probes together in
the labeling of intact A549 cells to obtain a comparable
photome list (Figure S5 and Supporting Information). Again,
a good distribution of cytosolic and membrane proteins was
obtained while re-extraction of the insoluble fraction did not
significantly enhance the overall coverage of membrane-
bound proteins (Figure S6).
With a comprehensive photome background list in hand,
we next analyzed the irradiation and concentration depend-
ence of DA photocrosslinking reactions. For this study, we
used the kinase inhibitor SP600125 equipped with a minimal
DA linker, similar as previously reported (Scheme S4).[4b]
Interestingly, while short irradiation times of 5 min and high
probe concentrations of 3 mm enhanced photome labeling,
longer irradiation times (20–30 min) as well as lower concen-
trations (125 nm) minimized background enrichment in this
setting (Figures S7 and S8). Although JNK, a known target of
SP600125, was not among the hits, concentration-dependent
competition with the unmodified drug revealed several
carboxylases that also exhibit ATP binding sites as putative
targets (Figure S9).
For an initial evaluation of the photome list, we re-
analyzed previous results obtained from A549 profiling with
a falcarinol diazirine probe.[8a] A comparison of the proteins
identified with the probe and those found in the photome
classified four out of seven falcarinol hits, including HMOX2
and CTSD, as photome targets (Figure S10). Importantly,
ALDH2, a biochemically confirmed target of falcarinol, stood
out from the photome proteins, which confirms the general
utility of this approach for unbiased target identification.
Finally, we sought to illustrate the applicability of the
photome inventory list in a target identification study with
isoquinoline sulfonamide H8 (Figure 3A), a widely applied
inhibitor of the cAMP-dependent protein kinase (PKA).[12]
H8 and its structurally related analogue H89 are gold
standards in PKA research,[12b] and AfBPP with this molecule
should reveal PKA as a validated positive control.
Based on the lessons learned from the photocrosslinker
comparisons described above, we equipped H8 with a short
aliphatic diazirine alkyne linker (Figure 3B), found to exhibit
minimal background labeling, at the terminal amine. Crys-
tallography data of PKA in complex with H-inhibitors further
indicate that chain extension at this position is tolerated.[13]
However, H8p might alternatively directly bind to the cAMP-
binding pocket of the regulatory subunit. The synthesis of this
probe was conducted through the coupling of the alkyl
diazirine carboxylic acid L4 (for the structure, see Scheme S1)
and N-(2-aminoethyl)isoquinoline-5-sulfonamide (2), which
was prepared according to published procedures,[14] in the
final step (Figure 3B).
A549 SILAC cells were treated with H8p and DMSO
according to established procedures. Proteomics workflow
and analysis revealed the enrichment of 30 proteins matching
our cut-off criteria (Figure 3C). These hits were classified into
three categories according to their enrichment in the photome
overlay (Figure 2D): 1) low confidence (five hits depicted in
red: log2(enrichment) in the photome > 1 and p < 0.05),
2) medium confidence (six hits depicted in orange: 0 <
log2(enrichment) ꢀ 1 and p < 0.05), 3) high confidence (19
remaining hits depicted in gray, which are not significantly
enriched in the photome). Interestingly, medium-and high-
Interestingly, reported abundances (Table S1) revealed
that proteins solely identified in one cell line, such as ECE1 in
HeLa cells, are less expressed in the other cell line, likely
explaining their cell-line-specific DA labeling. Importantly,
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 2. Background photoaffinity labeling of structurally diverse diazirines. A) Venn diagram of overlapping significantly enriched hits
(log2(enrichment)>1 and p<0.05) for all diazirine probes in live A549 cells. B) Venn diagram for the analysis in live HeLa cells. C) General
workflow to yield the combined diazirine photome. Four distinct diazirine proteome enrichments were analyzed individually by MaxQuant[16] and
subsequently merged (see the Methods Section in the Supporting Information). The output file was then statistically evaluated by a one-sample
Student’s t-test to gather mean enrichments and p-values over all valid quantifications. D) Volcano plot of the combined AfBPP results in A549
cells. To compile a master list of significant photome members, the overall significance level was adjusted by the method of Benjamini and
Hochberg (FDR=0.05). E) Volcano plot for the experiments with HeLa cells. F) List of top photome hits.
confidence proteins were predominantly linked to nucleotide
binding, such as ATP/cAMP, DNA, and polyadenine RNA
(Table S2). Furthermore, the regulatory subunit of PKA
(PRKAR2A) was identified.
In conclusion, our study has revealed the superior
performance of AA and DA compared to BP photocross-
linkers, which is in agreement with previous experiments
comparing crosslinking efficiencies.[5] A comparison of four
different DA scaffolds revealed a common set of false
positives. Proteins of high abundance and with a preference
for small-molecule binding belong to the most prominent hits.
Small linkers with aliphatic DAs led to the least binding while
aromatic substituents significantly increased the number of
hits. This information complements recent findings indicating
4
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 3. AfBPP of kinase inhibitor photoprobe H8p and photome-based confidence validation. A) Structures of two members of the H-series of
protein kinase A inhibitors: H8 and H89. B) Synthesis of H8p, a photoaffinity probe for H8. Reaction conditions: a) 1) SOCl2 (35.0 equiv), DMF,
reflux, 2 h; 2) NaHCO3 (1.00 equiv), H2O, 08C; 3) ethylenediamine (3.00 equiv), CH2Cl2, 08C!RT, 2 h, 52% over two steps; b) L4 (1.00 equiv),
HOBt (1.50 equiv), EDC (1.50 equiv), DIEA (2.00 equiv), DMF, RT, 16 h, 93%. C) Volcano plot representation of H8p AfBPP with 3 mm probe
(blow up of proteins located in the blue box shown on the right). Identified proteins were aligned with the diazirine photome (Figure 2D) and
classified as low-confidence (red; log2(photome enrichment)>1, p<0.05), medium-confidence (orange; 0<log2(photome enrichment)ꢀ1,
p<0.05), or high-confidence (gray, remaining proteins) hits. See the color code for a graphical representation of the hit positions in the photome.
D) Overlap of proteins present in the photome (red) and proteins identified in the H8p enrichment (blue). The regulatory subunit of PKA
(PRKAR2A) was thus identified. The compilation of the observed targets would have been impossible without our inventory list, once more
highlighting the importance of minimal photolinker controls for the exclusion of common non-specific hits.
Council (ERC starting grant to S.A.S.). M.S. was supported by
that branched photolinker scaffolds exhibit superior specific-
ity.[15] While it was possible to minimize off-target labeling
the Studienstiftung des deutschen Volkes. We thank Markus
Lakemeyer for helpful discussions and Katja Bꢀuml, Mona
Wolff, Daniel Nixdorf, Jacqueline Kissel, and Karolina Koch
within this isolated photocrosslinking study by reducing the
probe size and lowering logP (Table S3), in practice, photop-
for help with experiments and mass spectrometry. We thank
robe design is limited by the molecular scaffold of interest.
Megan H. Wright for critical proofreading of the manuscript.
Here, it becomes important to attach these groups at sites so
that the bioactivity and, if possible, the physicochemical
properties are not affected. By applying these design criteria
to a PKA inhibitor, we demonstrated successful target
Keywords: enzymes · photoaffinity labeling ·
photoreactive probes · protein profiling · proteomics
identification and the necessity for a photocrosslinker-specific
exclusion list. Thus we highly recommend the use of minimal
photocrosslinkers and their photome inventory list as an
important standard control for such experiments.
[1] a) E. Smith, I. Collins, Future Med. Chem. 2015, 7, 159 – 183;
b) P. P. Geurink, L. M. Prely, G. A. van der Marel, R. Bischoff,
H. S. Overkleeft, Top. Curr. Chem. 2012, 324, 85 – 113; c) D. J.
Acknowledgements
Lapinsky, D. S. Johnson, Future Med. Chem. 2015, 7, 2143 – 2171.
[2] a) A. Saghatelian, N. Jessani, A. Joseph, M. Humphrey, B. F.
We are grateful for funding from the Deutsche Forschungs-
gemeinschaft, CIPSM (to S.A.S.), and the European Research
Cravatt, Proc. Natl. Acad. Sci. USA 2004, 101, 10000 – 10005;
b) D. J. Lapinsky, Bioorg. Med. Chem. 2012, 20, 6237 – 6247.
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

Angewandte
Communications
Chemie
[3] a) L. Dubinsky, B. P. Krom, M. M. Meijler, Bioorg. Med. Chem.
2012, 20, 554 – 570; b) J. Das, Chem. Rev. 2011, 111, 4405 – 4417.
[4] a) A. L. MacKinnon, J. L. Garrison, R. S. Hegde, J. Taunton, J.
Am. Chem. Soc. 2007, 129, 14560 – 14561; b) Z. Li, P. Hao, L. Li,
C. Y. Tan, X. Cheng, G. Y. Chen, S. K. Sze, H. M. Shen, S. Q.
Yao, Angew. Chem. Int. Ed. 2013, 52, 8551 – 8556; Angew. Chem.
2013, 125, 8713 – 8718.
[5] a) K. Sakurai, S. Ozawa, R. Yamada, T. Yasui, S. Mizuno,
ChemBioChem 2014, 15, 1399 – 1403; b) J. T. Bush, L. J. Walport,
J. F. McGouran, I. K. H. Leung, G. Berridge, S. S. Van Berkel, A.
Basak, B. M. Kessler, C. J. Schofield, Chem. Sci. 2013, 4, 4115 –
4120.
[10] S. E. Ong, B. Blagoev, I. Kratchmarova, D. B. Kristensen, H.
Steen, A. Pandey, M. Mann, Mol. Cell. Proteomics 2002, 1, 376 –
386.
[11] A. Kawamura, D. M. Mihai, Methods Mol. Biol. 2012, 803, 65 –
75.
[12] a) M. Inagaki, S. Kawamoto, H. Itoh, M. Saitoh, M. Hagiwara, J.
Takahashi, H. Hidaka, Mol. Pharmacol. 1986, 29, 577 – 581; b) A.
Lochner, J. A. Moolman, Cardiovasc. Drug Rev. 2006, 24, 261 –
274.
[13] R. A. Engh, A. Girod, V. Kinzel, R. Huber, D. Bossemeyer, J.
Biol. Chem. 1996, 271, 26157 – 26164.
[14] A. Morikawa, T. Sone, T. Asano, J. Med. Chem. 1989, 32, 42 – 46.
[15] H. Park, J. Y. Koo, Y. V. Srikanth, S. Oh, J. Lee, J. Park, S. B.
Park, Chem. Commun. 2016, 52, 5828 – 5831.
[6] A. L. Mackinnon, J. Taunton, Curr. Protoc. Chem. Biol. 2009, 1,
55 – 73.
[7] J. Park, M. Koh, J. Y. Koo, S. Lee, S. B. Park, ACS Chem. Biol.
2016, 11, 44 – 52.
[16] MaxQuant enables high peptide identification rates, individu-
alized ppb range mass accuracies, and proteome-wide protein
quantification; see: J. Cox, M. Mann, Nat. Biotechnol. 2008, 26,
1367 – 1372.
[8] a) W. Heydenreuter, E. Kunold, S. A. Sieber, Chem. Commun.
2015, 51, 15784 – 15787; b) J. Eirich, S. Braig, L. Schyschka, P.
Servatius, J. Hoffmann, S. Hecht, S. Fulda, S. Zahler, I. Antes, U.
Kazmaier, S. A. Sieber, A. M. Vollmar, Angew. Chem. Int. Ed.
2014, 53, 12960 – 12965; Angew. Chem. 2014, 126, 13174 – 13179.
[9] a) C. W. Tornøe, C. Christensen, M. Meldal, J. Org. Chem. 2002,
67, 3057 – 3064; b) V. V. Rostovtsev, J. G. Green, V. V. Fokin,
K. B. Sharpless, Angew. Chem. Int. Ed. 2002, 41, 2596 – 2599;
Angew. Chem. 2002, 114, 2708 – 2711.
Manuscript received: June 20, 2016
Revised: September 15, 2016
Final Article published: && &&, &&&&
6
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
Photoaffinity Labeling
P. Kleiner, W. Heydenreuter, M. Stahl,
V. S. Korotkov,
S. A. Sieber*
&&&& — &&&&
A Whole Proteome Inventory of
Background Photocrosslinker Binding
In affinity-based protein profiling, small
molecules are equipped with a photo-
crosslinker for irreversible binding to
a protein target. The background protein
labeling observed with such photocross-
linkers was analyzed by gel-free quantita-
tive proteomics, and characteristic off-
targets were identified for each photo-
reactive group. This is of importance for
the reliable identification of biologically
relevant binders in photoprobe experi-
ments.
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7
These are not the final page numbers!