A comprehensive surface proteome analysis of myeloid leukemia cell lines for therapeutic antibody development

A comprehensive surface proteome analysis of myeloid leukemia cell lines for therapeutic antibody development

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1 Available online at www.sciencedirect.com ScienceDirect www.elsevier.com/locate/jprot A ...

2MB Sizes 3 Downloads 21 Views

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1

Available online at www.sciencedirect.com

ScienceDirect www.elsevier.com/locate/jprot

A comprehensive surface proteome analysis of myeloid leukemia cell lines for therapeutic antibody development Verena Strassbergera , Katrin L. Gutbrodta , Nikolaus Kralla , Christoph Roeslia,1 , Hitoshi Takizawac , Markus G. Manzc , Tim Fugmannb,⁎, Dario Neria,⁎⁎ a

ETH Zurich, Department of Chemistry and Applied Biosciences, Wolfgang-Pauli-Strasse 10, 8093 Zurich, Switzerland Philochem AG, Libernstrasse 3, 8112 Otelfingen, Switzerland c Division of Hematology, University Hospital Zurich, Raemistrasse 100, 8091 Zurich, Switzerland b

AR TIC LE I N FO

ABS TR ACT

Article history:

A detailed characterization of the cell surface proteome facilitates the identification of

Received 29 August 2013

target antigens, which can be used for the development of antibody-based therapeutics for

Accepted 11 January 2014

the treatment of hematological malignancies. We have performed cell surface biotinylation

Available online 30 January 2014

of five human myeloid leukemia cell lines and normal human granulocytes, which was used for mass spectrometric analysis and allowed the identification and label-free, relative

Keywords:

quantification of 320 membrane proteins. Several proteins exhibited a pronounced

Acute myeloid leukemia

difference in expression between leukemia cell lines and granulocytes. We focused our

Cell surface antigens

attention on CD166/ALCAM, as this protein was strongly up-regulated on all AML cell lines

Biotinylation

and AML blasts of some patients. A human monoclonal antibody specific to CD166

Proteomics

(named H8) was generated using phage display technology. H8 specifically recognized AML

Activated-leukocyte cell adhesion

cells in FACS analysis while demonstrating tumor targeting properties in vivo. After in vitro

molecule

screening of five potent cytotoxic agents, a duocarmycin derivative was used for the

Antibody–drug conjugates

preparation of an antibody–drug conjugate, which was able to kill AML cells in vitro with an IC50 of 8 nM. The presented atlas of surface proteins in myeloid leukemia provides an experimental basis for the choice of target antigens, which may be used for the development of anti-AML therapeutic antibodies. Biological significance The ability to discriminate between malignant and healthy, essential cells represents an important requirement for the development of armed antibodies for the therapy of hematological malignancies. Our proteomic study is, to our knowledge, the first large scale comparison of the accessible cell surface proteome of leukemia cells and normal blood cells, facilitating the choice of a suitable target for the treatment of acute myeloid leukemia (AML).

⁎ Corresponding author. Tel.: + 41 43 5448813; fax: + 41 43 5448809. ⁎⁎ Corresponding author. Tel.: + 41 44 6337401; fax: + 41 44 6331358. E-mail addresses: [email protected] (T. Fugmann), [email protected] (D. Neri). 1 Current address: Junior Research Group Biomarker Discovery, Deutsches Krebsforschungszentrum (DKFZ) and Heidelberg Institute for Stem Cell Technology and Experimental Medicine (HI-STEM gGmbH), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. 1874-3919/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jprot.2014.01.022

139

J O U RN A L OF P ROT EO M IC S 9 9 ( 2 01 4 ) 1 3 8 –1 51

An antibody drug conjugate was generated recognizing the CD166 antigen which was found to be strongly up-regulated in all AML cell lines and AML blasts of some patients. This antibody drug conjugate SIP(H8)-Duo might be further characterized in therapy experiments and might lead to a new targeted treatment option for AML. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Monoclonal antibodies are gaining importance for the treatment of hematological malignancies. In addition to approved products, such as Rituximab for the treatment of non-Hodgkin lymphoma (NHL) [1] and chronic lymphocytic leukemia (CLL), Ofatumumab for CLL and Alemtuzumab for CLL [2], several products are currently in advanced clinical development (e.g., Daratumumab for the treatment of multiple myeloma [3]) and we anticipate that a number of additional products will become routinely available for clinical use in the next few years. While traditionally intact antibodies have been used for therapeutic applications, there has been a recent trend towards arming antibodies with suitable payloads, such as drugs [4,5], radionuclides [6] or cytokines [7]. Indeed, Brentuximab vedotin is an antibody drug conjugate recently approved for the treatment of last-line post-transplant Hodgkin lymphoma and systemic anaplastic large cell lymphoma [8], while the radiolabeled anti-CD20 antibodies Ibritumomab tiuxetan and Iodine I 131 tositumomab have received marketing authorization for NHL treatment [8]. Moreover, the use of bispecific antibodies has exhibited promising results in NHL and ALL [9]. Antibodies recognize surface antigens on target cells. A comprehensive analysis of the surface proteome promises to be useful for the identification and validation of targets, which may be considered for the development of antibody-based therapeutics. Traditionally, the characterization of membrane proteins by conventional methods (such as two-dimensional gel electrophoresis) has been difficult, because these proteins tend to be insoluble in water and are not very abundant. However, the selective chemical modification of accessible proteins on the cell surface (e.g., by reaction of the proteins' primary amino groups [10,11] or of carbohydrate moieties [12] with reactive derivatives of biotin) may facilitate their enrichment and the subsequent mass spectrometry-based proteomic analysis of tryptic peptides. We have previously studied the surface proteome, either enriching accessible proteins by in vitro biotinylation of cell lines [11] or the in vivo biotinylation of the vasculature in healthy organs and at sites of disease [10]. Biotinylated proteins were purified on streptavidin resin, followed by on-resin digestion, yielding tryptic peptides which could be analyzed by mass spectrometry. While proteomics technologies continue to improve rapidly, the relative quantification of protein-derived peptides remains a challenge [13]. In this article, we present the results of a comparative surface proteome analysis of four acute myeloid leukemia cell lines (AML), one chronic myeloid leukemia (CML) cell line and of granulocytes isolated from normal human peripheral blood. Cell surface biotinylation, followed by capture on streptavidin resin, tryptic digestion and DeepQuanTR-assisted mass spectrometric analysis, allowed the identification and relative

quantification of 823 proteins. We studied in more detail CD166 or Activated Leukocyte Cell Adhesion Molecule (CD166/ ALCAM), as this protein was found to be strongly up-regulated in all AML cell lines, compared to CML and granulocyte controls. An anti-CD166 human monoclonal antibody (termed H8) was isolated, characterized and coupled to potent cytotoxic agents, in order to selectively kill AML cells. Finally, the expression of CD166 on human peripheral blood and bone marrow cells of AML patients and healthy donors was analyzed.

2. Materials and methods 2.1. Cell culture NB4 (ACC 207), THP1 (ACC 16), K562 (ACC 10), PLB985 (ACC 139) (all DSMZ) and HL60 (ATCC® CCL240™) cells were maintained in RPMI1640/L-Glutamine (Life Technologies, Inc., Carlsbad, CA), supplemented with 10% FBS (20% for HL60) (Life Technologies) and 1× antibiotic–antimycotic (Life Technologies) at 37 °C and 5% CO2. For the localized xenograft models 107 NB4 cells were subcutaneously injected into the right shoulder of 7 to 9 weeks old BALB/c nude mice (Charles River Laboratories, Sulzfeld, Germany). All animal experiments were performed on the basis of project license (42/2012) granted by the Veterinaeramt des Kantons Zuerich and approved by all participating institutions.

2.2. Purification of human polymorphonuclear leukocytes (PMN) From each of four different probands, 40 ml of normal human venous blood was collected with sodium citrate S-Monovette (Sarstedt, Nuembrecht, Germany) and processed and analyzed individually. Informed consent was obtained from each subject before blood collection. PMNs were purified using Polymorphprep™ (Axis-Shield PoC AS, Oslo, Norway) following the manufacturer's guidelines. Erythrocyte contamination was removed with 16.2 ml H2O for 20 s followed by 5.4 ml 0.6 M KCl and 33.2 ml PBS. At least 99.5% purity was reached as assessed by Wright Giemsa staining.

2.3. Cell surface biotinylation 3.4 × 107 cells were washed twice with ice cold PBS by 4 min centrifugation at 252 ×g and biotinylated for 5 min at room temperature on an orbital shaker with 6 ml 411 μM EZ-link sulfo-NHS-LC-biotin (Thermo Fisher Scientific, Inc., Rockford, IL) in PBS followed by addition of 25 μmol Tris–HCl (pH 7.4) and washing once with PBS. Cells were lysed in PBS containing 2% Nonidet P-40, 0.2% SDS, 10 mM EDTA by sonication for 9 s before and after incubation on ice for 30 min with repeated vortexing. The lysate was centrifuged at 16,100 ×g for 10 min at 4 °C and protein concentration of

140

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1

the supernatant was determined using the BCA Protein Assay Reagent Kit (Thermo Fisher Scientific).

2.4. Purification of biotinylated proteins from cell lysates Lysate corresponding to 2.4 × 107 cells was incubated with streptavidin resin (GE Healthcare, Chalfont St. Giles, UK) in 2% SDS for 2 h, and subsequently washed 3 × with buffer A (1% Nonidet P-40, 0.1% SDS in PBS pH 7.4), 2 × with buffer C (50 mM ammonium bicarbonate, pH 10), 2 × buffer B2 (2 M NaCl in PBS) and 8 × with digestion buffer (50 mM Tris–HCl, 1 mM CaCl2, pH 8.0). Finally, the resin was incubated in 220 μl digestion buffer containing 1.6 μg sequencing grade modified porcine trypsin (Promega Corporation, Madison, WI) for 16 h at 37 °C under constant agitation. Peptides were desalted, purified, and concentrated with OMIX C18 pipette tips (Agilent Technologies, Santa Clara, CA) following the manufacturer's guidelines, dried in a rotation vacuum concentrator and stored at − 20 °C.

2.5. Nano-capillary reverse-phase HPLC Dried tryptic peptides were dissolved in 22 μl 0.1% TFA in water of which 18 μl was separated by reverse-phase HPLC using EASY-nLC system (Thermo Fisher Scientific) with a ReproSil-Pur C18 AQ nano-RP column (Dr. Maisch GmbH, Germany) and a SunCollect microautosampler (SunChrom GmbH, Friedrichsdorf, Germany). Peptides were eluted with a gradient of 5–33% ACN over 62 min, mixed with four internal standard peptides and spotted on a blank MALDI target plate as described [14].

2.6. MALDI-TOF/TOF mass spectrometry MALDI-TOF/TOF analysis was conducted with the 4800 MALDI-TOF/TOF Analyzer (Life Technologies) as described [14]. Resulting spectra were processed and analyzed using the Global Protein Server Workstation (Version 3.6; AB SCIEX, using MASCOT version 2.1 (Matrix Science, London, UK)), for matching MS and MS/MS data against a database of all human and bovine in silico digested proteins (from UniProt homepage, contains 57,428 protein sequences). The bovine protein sequences account for possible contamination from FBS cell culture supplement. Therefore, proteins of bovine origin were not taken into further consideration. The following analysis settings were used for the identification of peptides and proteins: (i) precursor tolerance: 15 ppm, (ii) MS/MS fragment tolerance: 0.5 Da, (iii) maximal missed cleavages: one, (iv) one variable modification (oxidation of methionine) and (v) trypsin as enzyme. Peptides were considered correct calls when the individual confidence interval was greater than 95%.

2.7. Relative protein quantification and membrane protein annotation The DeepQuanTR software has been described previously [15] and the procedure to calculate the signal intensity of proteins is defined as published [14]. Briefly, after MS and MS/MS data acquisition, peak information was loaded into the DeepQuanTR software, normalized to the signal intensity of internal standard

peptides and aligned with other samples. Peptides were newly assembled to proteins using the corresponding feature integrated in DeepQuanTR. Briefly, a minimal protein list was established by annotating peptides to proteins and clustering of proteins sharing one or more peptides. If a protein was identified by at least one unique peptide, it was added to the protein list presented by DeepQuanTR. Groups of proteins sharing the same peptides (indistinguishable proteins) were ranked according to their sequence length to prevent random ranking. Only the top ranked protein of these is kept for the final list. Proteins were annotated to be associated with membranes, if one of three criteria was true: (i) The protein had at least one transmembrane helix as predicted by the TMHMM algorithm (version 2.0) [16], and/or the protein contained (ii) at least one transmembrane helix or (iii) a lipid anchor according to the UniProtKB/Swiss-Prot protein knowledge base.

2.8. Flow cytometry SIP antibodies were biotinylated with 20 fold molar excess of NHS-LC biotin (Thermo Fisher) in 5% DMSO in PBS for 1 h at room temperature and purified on PD10 columns (GE Healthcare). 5 × 105 cells were incubated in triplicate at 4 °C for 15 min in PBS containing 5% normal goat serum (NGS) (Millipore, Billeria, MA, USA) and 20 min in PBS containing 2% FBS (Life technologies, Paisley, UK), 15 μg/ml SIP(H8) or SIP(KSF) [17] as isotype control. After incubation for 20 min in PBS containing 2% FBS, 10 μg/ml Streptavidin-Alexa488 (Life technologies), cells were washed in PBS containing 2% FBS. Using directly labeled antibodies CD44-FITC (AbD Serotec) and the isotype control IgG1-FITC (AbD Serotec), 5 × 105 cells per sample were incubated in PBS containing 5% NGS for 10 min at 4 °C, 40 min at room temperature with antibodies diluted 1/150 in PBS containing 1% BSA and washed 3 × with PBS containing 1% BSA. 104 cells in PBS containing 1% BSA were analyzed on BD FACSCanto (Becton Dickinson) using FACSDiva software (Becton Dickinson) and FlowJo8.7.1 (TreeStar Ashland, TN) was used for data analysis. For each cell line, the log10 of the average shift in mean fluorescence intensity of the SIP(H8) or CD44 stained cells towards their isotype control was calculated and compared to the proteomics quant values calculated within the five cell lines. Primary human patient and stem cell donor samples were obtained based on cantonal ethical commission approval (Kantonale Ethik-Kommission Zürich, Nr. 2009-0062/1) and written donor informed consent at the Division of Hematology University Hospital Zurich. Mobilized peripheral blood and bone marrow CD34+ cells were obtained from healthy sibling hematopoietic stem cell donors, and following density gradient centrifugation to harvest mononuclear cells (MNCs), CD34+ cells were enriched using immunomagnetic beads according to the manufacturer's instructions (CD34 Microbead kit, Miltenyi Biotech, Germany). Bone marrow and peripheral blood leukemia samples were obtained from patients during medically indicated sampling procedures, and subsequently, MNCs were separated. In some cases, the MNCs were subjected to CD3+ and CD19+ cell depletion with CD3 and

J O U RN A L OF P ROT EO M IC S 9 9 ( 2 01 4 ) 1 3 8 –1 51

CD19 Microbeads (Miltenyi Biotech, Germany). The cells were frozen in 10% DMSO/90% FBS and stored at − 80 °C until the analysis. For the immunophenotyping, the frozen samples were thawed at 37 °C and stained in PBS with 2% FBS and 2 mM EDTA with the following antibodies: anti-human CD45 eFluor 450 (clone: HI30, eBioscience), anti-human CD34 PE-Cy7 (clone: 8G12, BD Biosciences), anti-human CD38 APC (clone: HIT2, BD Biosciences) and anti-human CD166 FITC (clone: 105902, R&D Systems) or isotype matched control (clone: 111711, R&D Systems). The stained samples were fixed with 2% PFA and analyzed on FACS Canto II (Becton Dickinson).

2.9. Selection of antibodies from the ETH-2-Gold library by phage display and SIP format generation Human monoclonal antibodies specific to CD166 antigen were isolated by two rounds of biopanning from the ETH-2-Gold antibody phage display library [18]. Antibody selections were performed on Immunotubes (Nunc, Roskilde, Denmark) coated with 50 μg/ml CD166(28-501)-His6 in PBS, as described [18]. Bacterial supernatants containing recombinant scFv antibody fragments were screened by ELISA [18]. Those with positive signal in ELISA were analyzed using a BIAcore 3000 instrument (GE Healthcare) and a CD166(28-501)-His6 coated CM5 sensor chip. Selected clones were sequenced using primers LMB3long and fdseqlong [18]. Antibodies in scFv format were converted into the SIP format by cloning VH and VL into pcDNA3.1 (Invitrogen) as described [19] and expressed in CHO-S cells [20]. Purified SIP antibodies were analyzed by SDS-PAGE. After size exclusion chromatography on Superdex 200 10/300 GL (GE Healthcare), fractions corresponding to the dimeric peak were analyzed by BIAcore on a CD166(28-501)-His6 coated CM5 sensor chip [21].

2.10. In vivo biodistribution of SIP(H8) SIP(H8) was modified with 25 equivalents of IRDye750 NHS ester (Licor, Lincoln, Nebraska, USA) for 1 h in PBS pH 7.4. After addition of 250 equivalents of Tris–HCl, the protein was purified using PD10 desalting column (GE Healthcare) and concentrated using Vivaspin 6 Centrifugal Concentrator (MWCO 10,000) (GE Healthcare). 100 μg was intravenously injected in triplicate into localized xenograft models, bearing a subcutaneous tumor on day 11 after injection of NB4 cells. At 12, 24, 29 and 48 h after injection, mice were imaged as described [22]. After 48 h mice were sacrificed, organs extracted, imaged, homogenized for quantification of signal intensities and imaged with a dilution series of IRDye750 NHS ester as described [22]. Tumor signal was corrected for tumor growth after injection. The IVIS (Xenogen, Caliper Life Sciences) settings were always λex = 745 nm, λem = 800 nm, exposure time = 1 s, f/stop = 2, medium binning.

2.11. In vitro cytotoxicity assays For the cytotoxicity assay [23], 15,000 cells per well were seeded in 100 μl complete medium containing different concentrations of drug, antibody drug conjugate (ADC) or just the respective solvent as control in triplicates. The ADC was dissolved in PBS, DM1 and DM4 [24] in 50% DMSO/water, CemCH2-SH [23] and Cem-CHO [25] in water and the duocarmycin derivative

141

[Supplementary information 4] (synthesis submitted elsewhere) in DMSO. Cells were incubated for 3 days at 37 °C in 5% CO2 and developed with Cell Titer 96 Aqueous One Solution Cell Proliferation Assay (Promega) according to the manufacturer's instructions. Measured data is corrected for pure medium absorbance and viability calculated in percent of drug treated sample over control.

2.12. Synthesis of ADCs The C-terminal cysteine residues of SIPs were reduced with 20 equivalents of tris(2-carboxyethyl) phosphine hydrochloride (TCEP·HCl) for 16 h at 4 °C and purified with HiTrap desalting columns (GE Healthcare) using degassed 5 mM EDTA in PBS. The duocarmycin derivative (Fig. 6d) (synthesis submitted elsewhere) was conjugated to the reduced SIP in 10% DMSO in PBS using 20 fold molar excess of drug over SIP monomer. After 1 h at room temperature under argon, the final ADC was purified with HiTrap Desalting columns (GE Healthcare) using PBS.

3. Results 3.1. Cell surface biotinylation and proteomic results We performed a cell surface biotinylation and comparative proteomic analysis of four AML cell lines (HL60, NB4, PLB985, THP1), one CML cell line (K562) and of normal human granulocytes (PMNs) using the procedure depicted in Fig. 1a. The use of the charged sulfo-NHS-LC-biotin generally allows for preferential biotinylation and subsequent streptavidin affinity purification of cell surface proteins. Label-free mass spectrometry and the following bioinformatic analysis are based on the DeepQuanTR software suite, which allows for the calculation of “peptide quant values”, leading to relative protein expression levels, expressed as “quant values” (Fig. 1a) [15]. A cluster analysis based on differences of mass spectrometric data revealed a high similarity between replicate analysis, while differences between cell lines and PMNs are more pronounced (Fig. 1b). In total, 823 proteins were identified with at least 2 peptides, including 320 proteins which could be annotated to the membrane, based on the TMHMM algorithm [16] and the UniProtKB/Swiss-Prot protein knowledge base. The percentage of putative membrane proteins in the sample groups ranged between 44 and 53% (Fig. 1c). The identification of non-membrane proteins may be a consequence of a partial penetration of the sulfo-NHS biotin reagent inside the cell, of the presence of dead cells in culture, or of surface exposure of putative intracellular proteins [11]. Putative membrane proteins were clustered into 8 clusters on the basis of quant values using the gene cluster 3.0 software [26]. Fig. 2 displays a selection of 56 proteins with representative examples of each cluster and presents a graphical protein quant display of relative protein abundance in the different specimens. A complete list of all identified proteins and protein quant values can be found in the Supplementary information 1. In total, 83 CD antigens were identified, including CD33 (Fig. 2), which has previously been used as target for the development of antibody– drug conjugates for the treatment of AML patients [27–29]. In our

142

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1

a HL60

THP1

NB4

PLB985

K562

PMN

Cell surface biotinylation with Sulfo-NHS-LC-Biotin

Cell lysis Streptavidin affinity purification On resin tryptic digestion RP-HPLC

Biotin

MALDI-TOF + MALDI-TOF/TOF

Peptide quants 50000 40000 30000 20000 10000 0

Protein quants HL60 THP1 NB4 PLB985 K562 PMN

HL60

b

PMN PMN PMN PMN HL60 HL60 K562

c K562

K562 THP1 THP1 THP1 NB4 NB4 NB4 PLB986 PLB986 PLB986

Membrane proteins per sample group

60% 50% 40% 30% 20% 10% 0%

Fig. 1 – Cell surface biotinylation and relative quantitative proteomic analysis of four AML cell lines (HL60, NB4, PLB985, THP1), one CML cell line (K562) and of normal human granulocytes (polymorphonuclear cells/PMN). (a) Schematic workflow of cell surface biotinylation using sulfo-NHS-LC-biotin, followed by the purification of biotinylated proteins on streptavidin resin in the presence of strong detergents, tryptic on-resin digestion and subsequent mass spectrometric characterization of resulting peptides. Using the DeepQuanTR software [15], internal standard peptides in each fraction are used for normalization of signal intensities, allowing for the robust relative quantification of peptide signal intensities, displayed as “peptide quant values”. The sum of peptide quant values from peptides belonging to the same protein is used to compute “protein quant values”, which can be graphically displayed as a color scale, ranging from bright red (down-regulated) to bright green (up-regulated) to visualize relative protein expression levels [15]. (b) Phylogenetic tree displaying the result of an unsupervised clustering of all specimens based on the differences in mass spectrometric data. (c) Percentage of putative membrane proteins in the sample groups. Proteins were annotated to the membrane if their UniProtKB/Swiss-Prot entry identified them to feature either at least one transmembrane helix or GPI-anchor, or if the TMHMM algorithm [16] predicted the protein to have at least one transmembrane helix.

study, CD33 was found to be strongly up-regulated in all leukemia cell lines, compared to granulocytes. Some proteins were found to be up-regulated in AML cells, compared to both CML cells and granulocytes (e.g., CD166, integrin alpha-4 and

embigin). Other proteins were up-regulated in CML cells (e.g., PROCR/CD201, TMEM2). Proteins up-regulated in granulocytes, compared to leukemia cell lines, included CD177, CD85b/LILRB2, CD85a/LILRB3 and CD11b/ITGAM (Fig. 2).

J O U RN A L OF P ROT EO M IC S 9 9 ( 2 01 4 ) 1 3 8 –1 51

143

Fig. 2 – Relative quantification of putative membrane proteins. Putative membrane proteins were relatively quantified and clustered with gene cluster 3.0 [26] into 8 clusters based on differences of mass spectrometric data using K-means method and Spearman's rank correlation. Representative proteins of each cluster are shown. Each square displays the protein quant value as color code (maximum = e5-fold up-regulation, green; minimum e5-fold down-regulation, red) representing the relative protein abundance in each sample.

In order to obtain an independent confirmation of the DeepQuanTR results, we performed a FACS analysis using antibodies specific to CD166 and to CD44 (Fig. 3). An excellent

agreement between experimental FACS data and protein quant values was observed (see for example the absence of FACS shift for CD44 in PLB985 cells, in relation to the red color

144

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1

a CD166

FACS

% of max

PLB985

HL60

K562

NB4

THP1

100

100

100

100

100

80

80

80

80

80

60

60

60

60

60

40

40

40

40

40

20

20

20

20

20

0 0 0 0 0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 FITC-A

Protein Quant Value

0.310

0.325

-1.648

0.680

0.332

b CD44

FACS

% of max

PLB985

HL60

NB4

K562

THP1

100

100

100

100

100

800

80

80

80

80

600

60

60

60

60

400

40

40

40

40

200

20

20

20

20

00 0 0 0 0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104

Protein Quant Value

FITC-A

-1.515

0.576

-0.393

1.203

0.129

Fig. 3 – Validation of the proteomics results by FACS. Proteomics data is represented by quant values. FACS staining with (a) anti-CD166 SIP(H8) or (b) anti-CD44 antibodies are presented as black lines, while FACS profiled with isotype control antibodies are depicted in gray.

of the three PLB replicates in the mass spectrometric analysis of Fig. 2, or the protein quant values depicted in Fig. 3). Calculation of the coefficient of correlation for CD166 (R2 = 0.90) and for CD44 (R2 = 0.65) confirmed a strong correlation of FACS and proteomics data.

3.2. Anti-CD166 antibody production and characterization We focused our attention on CD166 as a putative target for the development of anti-AML antibodies, since the antigen was found to be strongly up-regulated in all four AML cell lines. For antibody isolation, we expressed all five extracellular immunoglobulin domains of CD166 as recombinant protein (Supplementary information 2) and used it for selections with the antibody phage library ETH-2 Gold [18]. Antibody phage library technology [30] allows the facile isolation of fully human monoclonal antibodies, which are preferred for therapeutic applications, as they tend to be less immunogenic in humans. Clone H8, which was selected for further studies based on its strong ELISA signal and favorable BIAcore profile, was subcloned into the small immunoprotein (SIP) format and expressed in mammalian cells (Fig. 4). The SIP format features the scFv fragment at the N-terminal extremity of a human εCH4 domain of the secretory isoform S2 of human IgE [19]. This domain promotes the formation of homodimers that are further stabilized by disulfide bonds between the C-terminal cysteine residues, resulting in an about 80-kDa bivalent structure. Antibodies in SIP format display superior tumor targeting properties compared to IgG and scFv fragments, as

evidenced by quantitative biodistribution studies [19] and by radioimmunotherapy [31]. The C-terminal cysteine residue of each εCH4 domain can be chemically modified without loss of immunoreactivity or tumor-targeting performance allowing site specific modifications for the generation of ADCs [23]. SIP(H8) was purified to homogeneity and showed favorable biochemical properties in size exclusion chromatography (Fig. 4b), SDS-PAGE analysis (Fig. 4c), BIAcore analysis performed on antigen-coated microsensor chip (Fig. 4d) and FACS experiments on the five leukemia cell lines (Fig. 3a). The amino acid sequence of the H8 antibody can be found in Supplementary information 3. The in vivo targeting performance of IRDye750 labeled SIP(H8) was evaluated by a quantitative biodistribution study in mice bearing a subcutaneous NB4 tumor. Mice were imaged at different time points (12, 24, 29 and 48 h) after antibody injection and sacrificed at 48 h where tumor targeting compared to organs, except excretion related organs, could be demonstrated (Fig. 5).

3.3. Antibody–drug conjugates In view of the recent trend to develop ADCs for the treatment of acute leukemias, we performed a comparative analysis of five highly potent cytotoxic drugs, which could be coupled to antibodies for ADC generation. Duocarmycins are potent DNA alkylators and notable for their extreme cytotoxicity [32]. DM1 and DM4 are maytansinoids, which inhibit microtubule dynamics, leading to apoptosis

145

a anti-CD166 scFv(H8) VH

b

anti-CD166 SIP(H8) VH

VL

V

VL

VH

L

CH4 CH4

Absorption 280nm (mAU)

J O U RN A L OF P ROT EO M IC S 9 9 ( 2 01 4 ) 1 3 8 –1 51

25 20 15 10 5 0 0

kDa

c

10

20

30

Retention volume (ml)

1 2

d

250 – 130 – 100 – 70 –

250

RU

200

55 –

250 nM

150

125 nM

100 63 nM 31 nM 16 nM

50

35 – 25 –

0 0

100

15 – 10 –

200

Time (s)

Fig. 4 – Cloning and characterization of anti-CD166 SIP(H8). (a) Schematic representation of scFv(H8), consisting of a heavy chain (VH) and a light chain (VL) linked by a peptide linker. The H8 antibody was isolated from a phage library screened against the extracellular domain of CD166. The scFv antibody was reformatted into the SIP format by genetically fusing the scFv moiety to the N-terminal extremity of a human εCH4 domain of the secretory isoform S2 of human IgE leading to the formation of homodimers, further stabilized by a disulfide bond on the C-terminus [19]. (b) Size exclusion chromatography of SIP(H8) on a Superdex 200 10/300 GL column. The major peak at 13.5 ml corresponds to the molecular weight of the covalent homodimer (76 kDa). (c) SDS-PAGE of SIP(H8). Lane 1, non-reduced SIP(H8); lane 2, reduced SIP(H8). (d) Surface plasmon resonance measurements on an antigen coated microsensor chip with the covalent homodimer of SIP(H8), which had been purified by size exclusion chromatography.

[33]. DM1 has been used for an anti-CD33 ADC in clinical trials for the treatment of AML [29] and is the active drug in the recently approved ADC Trastuzumab emtansine [5]. Cemadotin,

a 12h

24h

29h

48h

a water soluble analog of dolastatin 15 also inhibits microtubule dynamics [34]. We have previously synthesized a thiol containing cemadotin analog (CemCH2-SH) for disulfide site-selective

Epi-fluorescence (counts)

c 80

4500 4000

75 8.0

3000 2500

b Organs 1 Tu

Li

Organs 2 Ki

Tu

Li

Organs 3 Ki

Tu

Li

Ki

Epi-fluorescence (counts) 3000 2800

% ID / g

3500

6.0 4.0 2.0

2600 Lu

He

In

Lu

He

In

Lu

He

Sp

Mu

Bl

Sp

Mu

Bl

Sp

Mu

Bl

2400 2200 2000

0.0

n/a n/a

Tu m o Lu r ng H ea Sp rt le M en us c B le lo o K d id n In ey te st in Li e ve r

In

Fig. 5 – In vivo imaging and biodistribution study. IRDye750 labeled SIP(H8) was intravenously injected in three mice bearing a subcutaneous tumor of NB4 cells on the right shoulder. (a) Imaging of mice at 12, 24, 29 and 48 h post injection. The representative images of mouse 2 are shown. The white arrow indicates the tumor. (b) Imaging of tumor (Tu) and organs liver (Li), kidney (Ki), intestine (In), lung (Lu), heart (He), spleen (Sp), muscle (Mu) and of blood (Bl) of all three mice 48 h after injection. (c) Signal intensity quantification of tissue homogenates of the three mice, 48 h after injection, displayed as the average with standard deviation in injected dose per gram of tissue (%ID/g). Signals in muscle and blood were below the detection limit (n/a).

146

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1

conjugation to an antibody [23] or containing an aldehyde moiety (Cem-CHO) allowing for thiazolidine linkage to an antibody [25]. In vitro cytotoxicity assays were performed on the four AML cell lines NB4, HL60, THP1 and PLB985. In all cell lines, the duocarmycin derivative exhibited the highest cytotoxic activity in the subnanomolar range, followed by DM4, Cem-CHO, DM1 and Cem-CH2-SH (Fig. 6). Due to this superior potency, we chose a duocarmycin derivative for the generation of an anti-CD166 antibody– drug conjugate, which could be suitable for the treatment of AML. The duocarmycin derivative for ADC preparation featured a linker, which contained both a carbamate moiety and a pyridyl-dithio moiety. The latter enabled the formation of mixed disulfides with reduced SIP(H8) (Fig. 7a,d). The purity of the preparation was confirmed by SDS-PAGE (Fig. 7b), while ESI-MS confirmed the correct molecular mass of the conjugate

NB4

3.4. CD166 expression in different stages of human hematopoietic cells In order to estimate the expression of CD166 in various human cell populations, peripheral blood cells of AML patients (P1–P4, Supplementary information 5) and healthy donors (C1–C3) were submitted to immunomagnetic enrichment procedures and analyzed by multicolor FACS (Fig. 8). Fractions containing mature hematopoietic cells (Fr. I, CD45hi) clearly showed low or no expression of CD166, which is in agreement with our proteomic findings on PMNs (Fig. 2). A high CD166 expression was detected in two (P1, P3) out of four patients in the fraction of mainly AML progenitors (Fr. IV,

100

100

80

80

60 40

60 40 20

20 0 10-11

THP1

120

Viability (%)

Viability (%)

120

(expected m/z: 38,724; found m/z: 38,727) (Fig. 7c). An in vitro cytotoxicity assay performed on HL60 cells demonstrated SIP(H8)-SS-Duocarmycin to be a potent antibody–drug conjugate, with an IC50 of 8.4 nM (Fig. 7e).

10-10

10-9

10-8

10-7

0 10-11

10-6

10-10

HL60

120

10-9

10-8

10-7

10-6

Concentration (M)

Concentration (M)

CemCH2-SH

PLB985

120

Cem-CHO DM1

100

100

80

80

DM4

Viability (%)

Viability (%)

Duo

60 40

40 20

20 0 10-11

60

10-10

10-9

10-8

10-7

10-6

0 10-11

10-10

10-9

10-8

10-7

10-6

Concentration (M)

Concentration (M)

IC50 (nM) CemCH2-SH

Cem-CHO

DM1

DM4

Duo

NB4

25

1.5

5.5

0.47

0.013

PLB985

68

4.1

9.2

2.1

0.15

THP1

61

6.1

6.2

1.5

0.20

HL60

75

5.4

7.9

1.8

0.22

Fig. 6 – In vitro screening of cytotoxic drugs. Cytotoxicity of Cem-CH2-SH, Cem-CHO, DM1, DM4 and a duocarmycin derivative (Duo) was measured by adding the compounds at various concentrations to the four AML cell lines NB4, THP1, HL60 and PLB985. After three days, cell viability was monitored with the Cell Titer 96 Aqueous One Solution Cell Proliferation Assay (Promega). Resulting IC50 values are depicted in the overview.

147

J O U RN A L OF P ROT EO M IC S 9 9 ( 2 01 4 ) 1 3 8 –1 51

a VH

VL

VL

VH

VH

VL

VL

VH

TCEP 20x/dimer CH4 CH4

HS

b

kDa 250 – 130 – 100 – 70 – 55 –

1 2 3 4

SH

VL

VH

CH4 CH4

Sephadex G-25 purification

CH4 CH4

Sephadex G-25 purification

20 eq Duocarmycin VL derivative/monomer VH RT, 1h, Argon

Drug - S S S

S - Drug

c 100

38727

100

% 0 900

1100

1300

1500

1700

%

76233

35 – 0

25 –

20000

30000

40000

50000

60000

70000

80000

m/z 120

d

e Viability (%)

100 80 60 40

IC50 = 8.4nM

20 0 10-12

10-11

10-10

10-9

10-8

10-7

Concentration (M) Fig. 7 – Synthesis and characterization of the antibody drug conjugate SIP(H8)-Duo. (a) Schematic representation of the conjugation strategy consisting of reduction of the C-terminal cysteine residues of SIP(H8) and, after removal of residual TCEP by gel filtration, chemical modification of the free thiols with the duocarmycin derivative leading to the formation of mixed disulfides. The resulting SIP(H8)-Duo is purified by gel filtration to remove free drug from the final product. (b) SDS-PAGE of SIP(H8) in the process of conjugation. Lane 1, non-reduced SIP(H8); lane 2, reduced SIP(H8); lane 3, SIP(H8)-Duo after 1 h reaction time; lane 4, SIP(H8)-Duo after purification. (c) ESI-MS spectrum of purified SIP(H8)-Duo (expected m/z: 38,724; found m/z: 38,727). The Peak of the m/z 76,233 corresponds to traces of unconjugated dimeric SIP(H8). (d) Chemical structure of the duocarmycin derivative used for conjugation to SIP(H8). (e) In vitro cytotoxicity of SIP(H8)-Duo with HL60 cells. The resulting IC50 value of 8.4 nM is depicted in the graph.

CD34+CD38+) and in the fraction in which the majority of leukemia stem cells (LSC) can be found (Fr. III, CD34+CD38−) [35,36]. However, two patients (P2, P4) showed no CD166 expression in these fractions. Peripheral blood samples of healthy donors showed CD166 expression in fractions of hematopoietic stem and earliest progenitor cells (Fr. III, CD34+CD38−) as well as progenitor cells (Fr. IV, CD34+CD38+). Whenever a bone marrow specimen was available from the same donor, multicolor FACS analysis was also performed on these bone marrow samples. This bone marrow FACS analysis showed a CD166 expression comparable to the one observed in peripheral blood samples (Supplementary information 6).

4. Discussion and conclusion The ability to discriminate between malignant and healthy, vitally critical cells represents an important requirement for the development of armed antibodies for the therapy of hematological malignancies. In order to facilitate the choice of a suitable target for the treatment of AML, we generated a comprehensive atlas of accessible proteins by cell surface biotinylation and label-free relative-quantitative proteomic analysis of four AML cell lines, one CML cell line and of granulocytes isolated from normal human peripheral blood.

148

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1

102

20

0

0 0

10 4

3.89

10 2

30.7

103

40

10

10

0 10

1.65

10 3

10 2

10

68 2

0 10

10

10 5

0 10 2

10 4

4

16.7

10 3

10

10

10

0 10

2

5

10

4

10 3

10 4

10 5

: CD38

1.96

10 5

0 10 2

10 4

28.1

10 3

10 2

16.6

103

40

0

0 10 3

10 4

10 5

0 10

10 5

4

10 3

0 10

46.5

10 2

0 0 10

10

10

10

: CD45

5

80

60 40

0 10 2

10 3

10 4

: ISO

10 5

10 4

40

0 10 2

10 5

10 3

10 4

10 3

10 5

10 4

10 5

: ISO

100

100

80

80

60 40

60 40 20 0

0 10

2

10 3

10 4

10 5

0 10

2

10 3

10 4

10 5

: ISO

100

100

80

80

60 40

60 40 20

0

: CD34

10 5

60

: ISO

49.8

10 4

0 10 3

20

0 10 2

10 3

20

0 10 2

10 5

102

0 4

10 4

42

103

40

0 3

10 3

104

20

2

2

2

: ISO

0

105

60

0 10

80

: CD34

80

% of Max

: CD34

AML PBMCs (PID257/SID516) Blast:32.0%

10

10 4

100

5

49.5

P4

10 3

: ISO

10 5

100

81.4

: CD38

10

2

10 4

20

0

: CD45

10 3

100

102

20

0 10 2

2

: ISO

104

60

40

0 0 10

10 5

105

63.3 4

10 4

10 5

60

0

10 3

10 4

20

: CD34

80 10

0 10 2

10 5

100

5

40

22

: CD38

AML PBMCs (PID72/SID130) Blast:55.5%

10 3

: ISO

% of Max

P3

: CD34

10

80

60

20

0 10 4

10 3

: ISO 100

102

0 10 3

0 10 2

10 5

: ISO

71.6

0

: CD45

10 4

80

10 5

103

40 20

0 10 2

10 4

104

60

10 2

40

0 10 3

0

105

80

10 3

10 3

10 5

20

: CD34

100

91.2

2

10 4

60

100

68.9 0 10

: ISO

% of Max

P2 AML PBMCs (PID28/SID49) Blast:60.0%

: CD34

: CD45 10

40

0 10 2

% of Max

0 10

80

60

20

0 5

10 3

: ISO 100

102

0 4

0 10 2

80

10 5

29.9

103

40

0 3

10 4

104

20

2

10

5

0

105

60

10 2

10

4

: ISO

100

75.1

3

0 10 2

10 5

3

20

: CD38

10

00 10

: ISO

% of Max

AML PBMCs (PID59/SID105) Blast:74.5%

: CD34

P1

10 3

40

100

51

10 5

20

2

0 10

102

10 4

60

: ISO

47.9

80 10

10

103

0

10 5

10

5

40

0 10 4

40

0 4

104 60

0

10 3

3

105

: CD45

60

: CD34

20

0 10 2

80

80

4

% of Max

: CD34

C3 Mobilized PB CD34+ (PBSC48)

10

5

100

97.9 10

10

4

: ISO

10 5

100

80

% of Max

10

: CD45

3

10 3

: ISO

% of Max

0 10

2

0 10 2

10 5

20

0 5

10 4

100

102

0 4

10 3

: ISO

104

60

0 3

0 0 10 2

10 5

105

20

2

10 4

% of Max

10 3

10 3

: CD34

80

% of Max

: CD34

C2

0 10 2

10 5

100

5

94.9

Mobilized PB CD34+ (PBSC30)

10 4

: CD38

10

10 3

: ISO

20

0

% of Max

0 10 2

: CD45

40

% of Max

10

5

% of Max

10

4

60

20

% of Max

10

3

40

Fr. III 36.8

% of Max

0 10

2

80

60

% of Max

102

80

% of Max

103

40

Fr. IV 55.5

100

% of Max

103

104

60

100

% of Max

Fr. I 99.6

: CD38

80 104

CD34+CD38+ (Fr. IV)

CD34+CD38- (Fr. III)

105

0.0151 % of Max

Buffy coat MNCs

: CD34

C1

CD45low (Fr. II)

100

Fr. II

% of Max

CD45hi (Fr. I)

Live singlet 105

0 0 10 2

10 3

10 4

: ISO

10 5

0 10 2

10 3

10 4

10 5

: ISO

Fig. 8 – Expression of CD166 in different stages of peripheral blood cells. Specimen from healthy donors, either mononuclear cells (MNCs, C1) or CD34+ enriched cells (C2, C3), were analyzed by multicolor FACS. Fraction I (CD45hi) represents mature peripheral blood cells while Fraction III (CD34+CD38−) represents mostly adult HSCs and Fraction IV (CD34+CD38+) represents normal progenitors. Specimen from AML patients (P1–P4) were also analyzed by multicolor FACS where Fraction I represents mature peripheral blood cells, Fraction III represents AML LSCs and Fraction IV represents leukemic progenitors as described in Kikushige et al. [35].

J O U RN A L OF P ROT EO M IC S 9 9 ( 2 01 4 ) 1 3 8 –1 51

We studied in more detail CD166, as this protein was found to be strongly up-regulated in all AML specimens, compared to the CML and granulocyte controls. An anti-CD166 human monoclonal antibody (termed H8) was isolated, characterized and coupled to a highly potent cytotoxic agent. The search for AML markers has previously heavily relied on transcriptomic data [37], but mRNA levels do not always reflect protein levels on the cell surface [38]. Similarly, the study of the whole proteome [39] in cell lysates provides only partial information about accessible proteins, which can be drugged using antibodies. Lee et al. [40] compared K562 to NB4 cells by cell surface biotinylation, but technologies available at that time allowed only for the identification of 25 membrane proteins. In a more recent study on leukemia stem cells, more than 1000 proteins were identified with at least one peptide of which 32% were annotated to the membrane after isolation of the membrane fraction by differential centrifugation [41]. In our study, using cell surface biotinylation, 823 proteins were identified with at least 2 peptides out of which 44–53% were predicted to be membrane proteins. The capture and relative quantification of N-linked glycopeptides combined with the use of variants of NHS-biotin [12] has been applied to the leukemia cell lines HL60 and NB4. However, the authors did not compare protein abundance versus normal granulocytes. Moreover, differential glycosylation may influence the quantification results obtained through this technique as cancer cells are known to feature aberrant protein glycosylation [42]. In our comparative proteomic study, we identified more than 100 proteins (e.g. CD33, CD166, integrin alpha-4 and embigin) being up-regulated in the AML cell lines compared to granulocytes which could be considered for AML targeting. Some proteins were also up-regulated specifically on the CML cells (e.g. TMEM2, CD201) and some proteins were specifically up-regulated on granulocytes (e.g. VSTM1, TREM1, CD177). To our knowledge, this is the first comparative study of the cell surface proteome featuring five leukemia cell lines and healthy blood cells, leading to the identification of several candidate antigens for the development of antibody-based therapeutics. One of the most up-regulated targets in Fig. 2 was the well-known antigen CD33, which is expressed by about 85–90% of AML patients [43,44] and has been used as target for the development of ADCs for the treatment of AML patients. Even though the ADC Gemtuzumab ozogamicin (Wyeth/Pfizer as Mylotarg) has been withdrawn from the market in 2010, some recent studies demonstrate clinical efficacy of Gemtuzumab ozogamicin combined with chemotherapeutics [28]. Seattle Genetics is developing a new ADC targeting CD33, with promising preclinical results. SGN-CD33A was shown to mediate complete regression of AML in mice at doses as low as 0.3 mg/kg [27]. We also found FLT3 on the surface of HL60, PLB985 and THP1 cells. Mutations in this gene can lead to constitutive signaling of this tyrosine kinase, contributing to factor-independent cell proliferation. Several FLT3 tyrosine kinase inhibitors are currently in clinical trials for the treatment of AML [45]. Integrin alpha-4/ITGA4/CD49d, which is targeted by the humanized antibody Natalizumab [46], IREM-1/CD300LF [47], which has been drugged by the chimeric monoclonal antibody D12 in preclinical studies, and markers which are expressed on leukemia stem cells (e.g. CD47, CD123, CD32) [48]

149

were also found to be up-regulated in AML cell lines (Fig. 2). This demonstrates that our study led to the identification of targets that already have gained or might gain clinical importance for the treatment of AML. We chose CD166 for the development of antibody drug conjugates, as this antigen was found to be strongly up-regulated in all AML specimens, compared to the CML and granulocyte controls. It had previously been shown that internalizing anti-CD166 antibodies can be selected from antibody phage display libraries [49,50] and that this antigen can be used for liposomal drug delivery [51]. In normal tissues CD166 is expressed in the spleen, placenta, liver [52], neurons [53], thymic epithelial cells, monocytes, activated T-cells and B-cells, but it is normally found only at very low levels in other blood cells [54]. Previous reports have documented the expression of CD166 in metastasizing human melanoma cell lines and it has been shown to be regulated in many cancers, including breast, colorectal, oral, ovarian, pancreatic, prostate and malignant melanomas [55]. A non-quantitative proteomic study of two AML patient blasts also identified CD166 as an expressed membrane protein [41]. In another study, an anti-CD166 scFv antibody fragment could be isolated by applying phage display selection on myeloid cell lines followed by a selection round on freshly isolated AML blasts [56]. We isolated a fully-human antibody fragment (named H8) using phage-display technology and reformatted it into the SIP format, since the C-terminal cysteine residue could be selectively chemically modified, leading to the preparation of an ADC with potent AML cell killing activity. The duocarmycin derivative used in our comparative cytotoxicity study was found to be more potent than the microtubule-targeting cemadotins and maytansinoids. As the pyrrolobenzodiazepine dimer of SGN-CD33A, the calicheamicin of Gemtuzumab ozogamicin or other chemotherapeutics with clinical relevance in AML (e.g. vosaroxin, decitabine, cytarabine and daunorubicin [57]), duocarmycins are typically classified as DNA damaging agents. They have been shown to bind to DNA and alkylate the nucleobase adenine at the N3 position [32]. As an alternative mechanism of action they have recently been proposed to bind and inhibit aldehyde dehydrogenases [58], which has been challenged by another recent publication [59]. We further analyzed the expression of CD166 on different stages of peripheral blood cells and bone marrow cells. Our data show expression of CD166 on normal hematopoietic stem and progenitor cells (HSPCs), indicating limits of the CD166 antigen for ADC development. However, also CD33, the so far most investigated target for ADC development in AML, can be found on normal HSPCs [60]. The FACS analysis also showed low or no CD166 expression on mature hematopoietic cells while some patients showed a clear CD166 expression on AML blasts, progenitors and leukemia stem cells (LSC). For these patients, SIP(H8)-SS-Duocarmycin or other similar ADCs, may be considered for preconditioning procedures and depletion of AML cells and HSPCs before allogeneic hematopoietic stem cell transplantation, while sparing mature hematopoietic cells. The comparative proteomic analysis of the relative abundance of accessible proteins on the surface of leukemia cells represents a starting point for the development of therapeutics based on intact antibodies or on armed antibodies. The experimental data were found to be in good agreement with

150

J O U RN A L OF P ROTE O M IC S 9 9 ( 2 01 4 ) 1 3 8 –15 1

FACS validation experiments and may represent the basis for antibody-based drug development programs for the targeting of human leukemias.

Conflict of interest The authors declare that there is no conflict of interest.

Acknowledgments Financial contributions from the Swiss National Science Foundation, the ETH Zürich, the Commission for Technology and Innovation (CTI) Switzerland, the Swiss Cancer League and the European Union (FP7 Project PRIAT) are gratefully acknowledged. This work was in part supported by the clinical research focus program of the University of Zurich. Verena Strassberger is a student of the Cancer Biology PhD Program of Life Science Zurich Graduate School. We would like to thank Dr. Giulio Casi (Philochem AG) for providing the cemadotin derivatives.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jprot.2014.01.022.

REFERENCES [1] Griffin MM, Morley N. Rituximab in the treatment of non-Hodgkin's lymphoma — a critical evaluation of randomized controlled trials. Expert Opin Biol Ther 2013;13:803–11. [2] Stephens DM, Byrd JC. Improving the treatment outcome of patients with chronic lymphocytic leukemia through targeted antibody therapy. Hematol Oncol Clin North Am 2013;27:303–27. [3] de Weers M, Tai YT, van der Veer MS, Bakker JM, Vink T, Jacobs DC, et al. Daratumumab, a novel therapeutic human CD38 monoclonal antibody, induces killing of multiple myeloma and other hematological tumors. J Immunol 2011;186:1840–8. [4] Sievers EL, Senter PD. Antibody–drug conjugates in cancer therapy. Annu Rev Med 2013;64:15–29. [5] Teicher BA, Chari RV. Antibody conjugate therapeutics: challenges and potential. Clin Cancer Res 2011;17:6389–97. [6] Pouget JP, Navarro-Teulon I, Bardies M, Chouin N, Cartron G, Pelegrin A, et al. Clinical radioimmunotherapy — the role of radiobiology. Nat Rev Clin Oncol 2011;8:720–34. [7] Pasche N, Neri D. Immunocytokines: a novel class of potent armed antibodies. Drug Discov Today 2012;17:583–90. [8] Garnock-Jones KP. Brentuximab vedotin: a review of its use in patients with Hodgkin lymphoma and systemic anaplastic large cell lymphoma following previous treatment failure. Drugs 2013;73:371–81. [9] Nagorsen D, Kufer P, Baeuerle PA, Bargou R. Blinatumomab: a historical perspective. Pharmacol Ther 2012;136:334–42. [10] Schliemann C, Roesli C, Kamada H, Borgia B, Fugmann T, Klapper W, et al. In vivo biotinylation of the vasculature in B-cell lymphoma identifies BST-2 as a target for antibody-based therapy. Blood 2010;115:736–44.

[11] Roesli C, Borgia B, Schliemann C, Gunthert M, Wunderli-Allenspach H, Giavazzi R, et al. Comparative analysis of the membrane proteome of closely related metastatic and nonmetastatic tumor cells. Cancer Res 2009;69:5406–14. [12] Hofmann A, Gerrits B, Schmidt A, Bock T, Bausch-Fluck D, Aebersold R, et al. Proteomic cell surface phenotyping of differentiating acute myeloid leukemia cells. Blood 2010;116: e26–34. [13] Nahnsen S, Bielow C, Reinert K, Kohlbacher O. Tools for label-free peptide quantification. Mol Cell Proteomics 2013;12:549–56. [14] Roesli C, Fugmann T, Borgia B, Schliemann C, Neri D, Jucker M. The accessible cerebral vascular proteome in a mouse model of cerebral beta-amyloidosis. J Proteomics 2011;74:539–46. [15] Fugmann T, Neri D, Roesli C. DeepQuanTR: MALDI-MS-based label-free quantification of proteins in complex biological samples. Proteomics 2010;10:2631–43. [16] Krogh A, Larsson B, von Heijne G, Sonnhammer EL. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J Mol Biol 2001;305:567–80. [17] Frey K, Fiechter M, Schwager K, Belloni B, Barysch MJ, Neri D, et al. Different patterns of fibronectin and tenascin-C splice variants expression in primary and metastatic melanoma lesions. Exp Dermatol 2011;20:685–8. [18] Silacci M, Brack S, Schirru G, Marlind J, Ettorre A, Merlo A, et al. Design, construction, and characterization of a large synthetic human antibody phage display library. Proteomics 2005;5:2340–50. [19] Borsi L, Balza E, Bestagno M, Castellani P, Carnemolla B, Biro A, et al. Selective targeting of tumoral vasculature: comparison of different formats of an antibody (L19) to the ED-B domain of fibronectin. Int J Cancer 2002;102:75–85. [20] Pasche N, Woytschak J, Wulhfard S, Villa A, Frey K, Neri D. Cloning and characterization of novel tumor-targeting immunocytokines based on murine IL7. J Biotechnol 2011;154:84–92. [21] Villa A, Trachsel E, Kaspar M, Schliemann C, Sommavilla R, Rybak JN, et al. A high-affinity human monoclonal antibody specific to the alternatively spliced EDA domain of fibronectin efficiently targets tumor neo-vasculature in vivo. Int J Cancer 2008;122:2405–13. [22] Steiner M, Gutbrodt K, Krall N, Neri D. Tumor-targeting antibody–anticalin fusion proteins for in vivo pretargeting applications. Bioconjug Chem 2013;24:234–41. [23] Bernardes GJ, Casi G, Trussel S, Hartmann I, Schwager K, Scheuermann J, et al. A traceless vascular-targeting antibody–drug conjugate for cancer therapy. Angew Chem 2012;51:941–4. [24] Widdison WC, Wilhelm SD, Cavanagh EE, Whiteman KR, Leece BA, Kovtun Y, et al. Semisynthetic maytansine analogues for the targeted treatment of cancer. J Med Chem 2006;49:4392–408. [25] Casi G, Huguenin-Dezot N, Zuberbuhler K, Scheuermann J, Neri D. Site-specific traceless coupling of potent cytotoxic drugs to recombinant antibodies for pharmacodelivery. J Am Chem Soc 2012;134:5887–92. [26] de Hoon MJ, Imoto S, Nolan J, Miyano S. Open source clustering software. Bioinformatics 2004;20:1453–4. [27] Sutherland MSK, Walter RB, Jeffrey SC, Burke PJ, Yu C, Harrington KH, et al. SGN-CD33A: A Novel CD33-Directed Antibody–Drug Conjugate, Utilizing Pyrrolobenzodiazepine Dimers, Demonstrates Preclinical Antitumor Activity Against Multi-Drug Resistant Human AML 54th ASH Annual Meeting and Exposition; 2012. [28] Burnett AK, Hills RK, Hunter AE, Milligan D, Kell WJ, Wheatley K, et al. The addition of gemtuzumab ozogamicin to low-dose

J O U RN A L OF P ROT EO M IC S 9 9 ( 2 01 4 ) 1 3 8 –1 51

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

Ara-C improves remission rate but does not significantly prolong survival in older patients with acute myeloid leukaemia: results from the LRF AML14 and NCRI AML16 pick-a-winner comparison. Leukemia 2013;27:75–81. Lapusan S, Vidriales MB, Thomas X, de Botton S, Vekhoff A, Tang R, et al. Phase I studies of AVE9633, an anti-CD33 antibody–maytansinoid conjugate, in adult patients with relapsed/refractory acute myeloid leukemia. Invest New Drugs 2012;30:1121–31. McCafferty J, Griffiths AD, Winter G, Chiswell DJ. Phage antibodies: filamentous phage displaying antibody variable domains. Nature 1990;348:552–4. Berndorff D, Borkowski S, Sieger S, Rother A, Friebe M, Viti F, et al. Radioimmunotherapy of solid tumors by targeting extra domain B fibronectin: identification of the best-suited radioimmunoconjugate. Clin Cancer Res 2005;11:7053s–63s. Boger DL, Johnson DS. CC-1065 and the duocarmycins: unraveling the keys to a new class of naturally derived DNA alkylating agents. Proc Natl Acad Sci U S A 1995;92:3642–9. Lopus M, Oroudjev E, Wilson L, Wilhelm S, Widdison W, Chari R, et al. Maytansine and cellular metabolites of antibody–maytansinoid conjugates strongly suppress microtubule dynamics by binding to microtubules. Mol Cancer Ther 2010;9:2689–99. de Arruda M, Cocchiaro CA, Nelson CM, Grinnell CM, Janssen B, Haupt A, et al. LU103793 (NSC D-669356): a synthetic peptide that interacts with microtubules and inhibits mitosis. Cancer Res 1995;55:3085–92. Kikushige Y, Shima T, Takayanagi S, Urata S, Miyamoto T, Iwasaki H, et al. TIM-3 is a promising target to selectively kill acute myeloid leukemia stem cells. Cell Stem Cell 2010;7:708–17. Eppert K, Takenaka K, Lechman ER, Waldron L, Nilsson B, van Galen P, et al. Stem cell gene expression programs influence clinical outcome in human leukemia. Nat Med 2011;17:1086–93. Valk PJ, Verhaak RG, Beijen MA, Erpelinck CA, Barjesteh van Waalwijk van Doorn-Khosrovani S, Boer JM, et al. Prognostically useful gene-expression profiles in acute myeloid leukemia. N Engl J Med 2004;350:1617–28. Gygi SP, Rochon Y, Franza BR, Aebersold R. Correlation between protein and mRNA abundance in yeast. Mol Cell Biol 1999;19:1720–30. Luczak M, Kazmierczak M, Handschuh L, Lewandowski K, Komarnicki M, Figlerowicz M. Comparative proteome analysis of acute myeloid leukemia with and without maturation. J Proteomics 2012;75:5734–48. Lee SJ, Kim KH, Park JS, Jung JW, Kim YH, Kim SK, et al. Comparative analysis of cell surface proteins in chronic and acute leukemia cell lines. Biochem Biophys Res Commun 2007;357:620–6. Bonardi F, Fusetti F, Deelen P, van Gosliga D, Vellenga E, Schuringa JJ. A proteomics and transcriptomics approach to identify leukemic stem cell (LSC) markers. Mol Cell Proteomics 2013;12:626–37. Ruhaak LR, Miyamoto S, Lebrilla CB. Developments in the identification of glycan biomarkers for the detection of cancer. Mol Cell Proteomics 2013;12:846–55. Dinndorf PA, Andrews RG, Benjamin D, Ridgway D, Wolff L, Bernstein ID. Expression of normal myeloid-associated antigens by acute leukemia cells. Blood 1986;67:1048–53.

151

[44] Griffin JD, Linch D, Sabbath K, Larcom P, Schlossman SF. A monoclonal antibody reactive with normal and leukemic human myeloid progenitor cells. Leuk Res 1984;8:521–34. [45] Kayser S, Levis MJ. Flt3 tyrosine kinase inhibitors in acute myeloid leukemia: clinical implications and limitations. Leukemia & lymphoma 2014;55(2):243–55. http://dx.doi.org/ 10.3109/10428194.2013.800198. [46] Steinman L. The discovery of natalizumab, a potent therapeutic for multiple sclerosis. J Cell Biol 2012;199:413–6. [47] Korver W, Zhao X, Singh S, Pardoux C, Zhao J, Guzman ML, et al. Monoclonal antibodies against IREM-1: potential for targeted therapy of AML. Leukemia 2009;23:1587–97. [48] Majeti R. Monoclonal antibody therapy directed against human acute myeloid leukemia stem cells. Oncogene 2011;30:1009–19. [49] Piazza T, Cha E, Bongarzone I, Canevari S, Bolognesi A, Polito L, et al. Internalization and recycling of ALCAM/CD166 detected by a fully human single-chain recombinant antibody. J Cell Sci 2005;118:1515–25. [50] Liu B, Conrad F, Roth A, Drummond DC, Simko JP, Marks JD. Recombinant full-length human IgG1s targeting hormone-refractory prostate cancer. J Mol Med 2007;85:1113–23. [51] Roth A, Drummond DC, Conrad F, Hayes ME, Kirpotin DB, Benz CC, et al. Anti-CD166 single chain antibody-mediated intracellular delivery of liposomal drugs to prostate cancer cells. Mol Cancer Ther 2007;6:2737–46. [52] Degen WG, van Kempen LC, Gijzen EG, van Groningen JJ, van Kooyk Y, Bloemers HP, et al. MEMD, a new cell adhesion molecule in metastasizing human melanoma cell lines, is identical to ALCAM (activated leukocyte cell adhesion molecule). Am J Pathol 1998;152:805–13. [53] Patel DD, Wee SF, Whichard LP, Bowen MA, Pesando JM, Aruffo A, et al. Identification and characterization of a 100-kD ligand for CD6 on human thymic epithelial cells. J Exp Med 1995;181:1563–8. [54] Bowen MA, Patel DD, Li X, Modrell B, Malacko AR, Wang WC, et al. Cloning, mapping, and characterization of activated leukocyte-cell adhesion molecule (ALCAM), a CD6 ligand. J Exp Med 1995;181:2213–20. [55] Weidle UH, Eggle D, Klostermann S, Swart GW. ALCAM/ CD166: cancer-related issues. Cancer Genomics Proteomics 2010;7:231–43. [56] Geuijen CA, Bijl N, Smit RC, Cox F, Throsby M, Visser TJ, et al. A proteomic approach to tumour target identification using phage display, affinity purification and mass spectrometry. Eur J Cancer 2005;41:178–87. [57] Roboz GJ. Novel approaches to the treatment of acute myeloid leukemia. Hematology Am Soc Hematol Educ Program 2011;2011:43–50. [58] Wirth T, Schmuck K, Tietze LF, Sieber SA. Duocarmycin analogues target aldehyde dehydrogenase 1 in lung cancer cells. Angew Chem 2012;51:2874–7. [59] Tercel M, McManaway SP, Leung E, Liyanage HD, Lu GL, Pruijn FB. The cytotoxicity of duocarmycin analogues is mediated through alkylation of DNA, not aldehyde dehydrogenase 1: a comment. Angew Chem 2013;52:5442–6. [60] Taussig DC, Pearce DJ, Simpson C, Rohatiner AZ, Lister TA, Kelly G, et al. Hematopoietic stem cells express multiple myeloid markers: implications for the origin and targeted therapy of acute myeloid leukemia. Blood 2005;106:4086–92.