lip-quant, an automated chemoproteomic approach … · depleted tumor models in pd-1 treatment ....
TRANSCRIPT
INTRODUCTION
RESULTS
CONCLUSIONS
• A multidisciplinary approach combining genetic (e.g. CRISPR-based screening), biochemical (e.g. CETSA) and biophysical techniques is needed to meet this challenge
• We present a novel approach of the target ID tool kit based on limited proteolysis (LiP) and quantitative mass spectrometry (MS)
LIP-QUANT, AN AUTOMATED CHEMOPROTEOMIC APPROACH TO IDENTIFY DRUG TARGETS IN COMPLEX PROTEOMES
Nigel Beaton1, Ilaria Piazza², Roland Bruderer¹, Yuehan Feng¹, Paola Picotti², Lukas Reiter1
1) Biognosys AG, Schlieren (Zurich), Switzerland 2) Institute of Molecular Systems Biology, ETH Zurich, Zurich, Switzerland
• Target identification plays a key role in both target-based drug discovery and phenotypic drug discovery
• In order to refine a promising drug candidate, extensive knowledge of its target protein(s), including undesirable off-target binding events, is essential
Figure 1: Machine Learning-based Chemoproteomic Pipeline for Small Molecule Target Deconvolution Mechanically sheared cell/tissue lysate is incubated with compound at multiple concentrations (left panel) followed by LiP and
Figure 4: Peptide Mapping on MEK1 Protein Mapping of three LiP-peptides (in blue, denoted with first and last amino acid residue) to the previously reported crystal structure (Robarge et al. 2014) of human MEK1 bound to selumetinib (in orange).
Yuehan Feng, PhDBusiness Development Manager
• The peptide-centric approach enables estimation of relative binding affinities and mapping of potential binding sites
• LiP-Quant is an addition to the target deconvolution toolbox that enables the probing of compound-target interactions, off-target binding, and can potentially provide crucial information regarding binding site prediction
• We present an integrated machine learning-based chemoproteomic workflow which can be applied to identify drug targets in complex proteomes
• LiP exploits orthogonal biophysical principles of compound binding - conformational changes and steric hindrance - for target deconvolution
Figure 2: Global Target Deconvolution for Kinase Inhibitors based on Ranking of LiP-peptides In two 7-series dose response experiments in HeLa lysate, the targets of a highly specific kinase inhibitor, selumetinib and a broad specificity inhibitor, staurosporine are identified. LiP scores of all ranked peptides (q-value < 0.01 and absolute log2 fold change > 0.46) are shown where peptides from kinases and MEK1/2 are colored in blue and red respectively.
Figure 3: Dose Response Curves for High-ranking LiP Peptides of MAP2K1 (Target of Selumetinib) and NQO2 (Off Target)
subsequent trypsin digestion. Resulting peptides are analyzed via data-independent acquisition mass spectrometry (DIA-MS). A machine learning-based model is employed to discern features indicative of drug binding (middle panel) and an integrated LiP score is assigned to each
peptide. Global LiP score distribution (right panel) for known target and non-target proteins shows clear separation (in positive control training data sets). LiP score ranks potential candidates in a target deconvolution experiment without bias.
INTRODUCTION
RESULTS
CONCLUSIONS
• Characterization of protein binding site(s) and induced conformational changes can enable better compound design
• We present a novel proteomics-based technique termed high resolution LiP (HR-LiP) that provides peptide-specific structural information for compound targets
LIMITED PROTEOLYSIS AS A NOVEL APPROACH TO PROTEIN STRUCTURAL ANALYSIS
Nigel Beaton1, Jagat Adhikari², Roland Bruderer¹, Ron Tomlinson², Yuehan Feng¹, Iván Cornella-Taracido² & Lukas Reiter¹
1) Biognosys AG, Wagistrasse 21, 8952 Schlieren (Zurich), Switzerland 2) Cedilla Therapeutics, 38 Sidney Street, 02139, Cambridge, USA
• Limited proteolysis (LiP) is a recently developed peptide-centric target deconvolution approach that can identify protein targets in a complex lysate
• LiP’s peptide-based identification of target proteins also enables a level of granularity often missing from other common target deconvolution techniques, including binding site approximation
Figure 1: HR-LiP Development with Calmodulin. (A) Calmodulin protein coverage in HR-LiP. (B) Peptides with high dose response correlation (> 0.6) upon addition of Ca2+ (n = 3). (C) Calmodulin peptide localization from (b) in red. (D) Mapping of peptides with high dose response correlation upon addition of CamKII peptide (blue) to Ca2+-bound calmodulin (n = 3) in red.
Figure 2: HR-LiP for Analysis of Compound Binding Sites.(A) Two compounds (BI-3802, left and JQ1, right) tested using HR-LiP. (B) Peptides regulated by binding in BCL6 (BI-3802) and BRD4 (JQ1). (C) BCL6 peptide mapping from (b) for BI-3802. (D) Blue and yellow peptides from (c) overlap with published hydrogen-deuterium exchange (HDX) data. (E) Dose response curve of top BRD4 peptide from (b). (F) Binding site of JQ1 in BRD4. (G) Peptide from (e) mapped to BRD4 (red). Amino acids (blue) from (f) highlighted.
Nigel Beaton, PhDSenior Scientist
• Model system work in calmodulin demonstrated that HR-LiP can approximate Ca2+ binding sites
• HR-LiP accurately predicts the binding site of well-characterized compounds (JQ1 and BI-3802), as well as additional sites of regulation (BI-3802) confirmed by HDX
• The ability to accurately map compound binding sites and conformational changes induced by compound binding remain a high priority in drug development
• HR-LiP is an extension of LiP-Quant that aims to exploit peptide-specific information to predict compound binding sites, as well as sites of conformational regulation
A
B
C
D
Stripped Sequence Protein R2
EAFSLFCALM1
0.704YEEFVQMMTAK 0.685SAAELR 0.607
ACompound Target Stripped Sequence R2
BI-3082 BCL6
TVLMACSGLFYSIFTDQLKR 0.763SRDILTDVVIVVSR 0.669EGNIMAVMATAMYLQMEHVVDTCR 0.638LTDVVIVVSR 0.616
JQ1 BRD4 FQQPVDAVK 0.895ETGTAKPGVS 0.699
B
D
Modified from Kerres et al., 2017
C
E F G
Modified from Muller et al., 2011
INTRODUCTION
RESULTS
CONCLUSIONS
IFN- γNetwork
MC38 aPD-1, aCD4MC38 aPD-1
DEEP PROTEOMIC PROFILING OF CD4+ AND CD8+ T CELL-DEPLETED TUMOR MODELS IN PD-1 TREATMENT
Jan Muntel1, Ying Jin2, Haochen Yu1, Nicholas Dupuis1, Yongli Shan2, Annie X. An2, Wubin Qian2, Davy Ouyang2, Kristina Beeler1, Roland Bruderer1
1) Biognosys AG, Wagistrasse 21, 8952 Schlieren (Zurich), Switzerland 2) Crown Bioscience Inc., San Diego (CA), USA
0500
100015002000250030003500
0 5 10 15 20 25
Tum
or V
olum
e (m
m³)
Study Days
MC38 PBSMC38 aPD-1MC38 aPD-1, aCD8MC38 aPD-1, aCD4
Granzyme B Perforin ZAP70 IDO1
PD-L1 LAG3 VISTA NOS2
Relative abundance (z-score)
PBSaPD-1aPD-1, aCD8aPD-1, aCD4
Hepa1-6MC38
TUMOR MODEL
TREATMENT
Immunotherapies targeting the PD-1/PD-L1 axis have been shown to be effective in only ~20% of cancer patients and the mechanism of action (MOA) underlying the differences between responders and non-responders remains poorly understood. Immunophenotyping of patient samples play important roles in understanding MOAs, however a clear understanding of the
heterogeneous responses remains elusive due to the intrinsic complexity of tumor immunity. It is therefore critical to understand the roles of different lineages of immune cells in mediating PD-1 response, while gaining an overall understanding of the tumor micro-environment (TME) . To address this, tumor-bearing syngenic mice were treated with an anti-PD-1 antibody in combination
Figure 1. Schematic Representation of the Mass Spectrometric WorkflowProteins are extracted from 1mg of FFPE tissue for each sample and processed into tryptic peptides. Quantification was performed with DIA-MS/HRM HighD workflow. Data analysis was conducted using SpectronautTM software.
Figure 3. Functional Analysis Identify IFN-γ Up-regulation Associated with CD4 Depletion in MC38 (A) In MC 38, PLS-DA analysis distinguishes between groups treated with anti-PD1 versus anti-PD1 and anti-CD4 (B) 12 out of the 25 top candidate proteins identified from the PLS-DA analysis are involved in IFN-γ signaling cascade (C) Upstream regulator analysis shows that the proteomic data is mostly consistent with the
Figure 2. Proteomics Data Recapitulates Efficacy Data (A) 9’248 proteins are quantified. Unsupervised hierarchical clustering clearly distinguishes two mouse strains but not different treatment groups(B) In MC38 immuno-oncology (IO) model, key protein markers demonstrate treatment responses consistent with tumor suppression efficacy.
Jan Muntel, PhDSenior Scientist
with targeted CD4+ or CD8+ T cell depletion with the purpose of understanding how these lineages impacted anti-PD-1 efficacy. Tumor tissues were subsequently investigated with an unbiased proteomics workflow based on data-independent acquisition (DIA) mass spectrometry to provide insights into TME in different treatment contexts.
A B
C
D
BA
• Syngenic mouse model show differential response to PD1 blockade and CD4+, CD8+ T cell depletion
• Hyper Reaction Monitoring Mass Spectrometry (HRMTM-MS) discovery proteomics quantified over 9’000 Proteins
• Functional analysis suggests that CD4 depletion boosts tumor suppression under PD1 blockade through up-regulation of IFN-γ signaling cascade
predicted regulations of proteins in when IFN-γ is unregulated (D) The proteomic data on key IO markers correlate with the RNASeq data from the same tumor samples.
MC38 PBS MC38 aPD-1 MC38 aPD-1, aCD8 MC38 aPD-1, aCD4
INTRODUCTION
RESULTS
CONCLUSIONS
QUANTITATIVE PROTEOMICS REVEALS NOVEL IMMUNOMODULATORY PATHWAYS OF RESISTANCE TO PARP THERAPY
Jakob Vowinckel1, Yuehan Feng1, Dimitra Georgopoulou2, Alejandra Bruna2, Tobias Treiber1, Abigail Shea2, Giulia Lerda2, Martin O’Reilly2, Kristina Beeler1, Carlos Caldas2
1) Biognosys AG, Schlieren (Zurich), Switzerland 2) Department of Oncology and Cancer Research UK Cambridge Institute, Li Ka Shing Centre, University of Cambridge, Cambridge, UK
Pharmacological inhibition of PARP results in the specific killing of BRCA1/2 deficient tumor cells due to a synthetic lethal interaction between the concomitant impairment in homologous recombination and the DNA damage response.
Despite the success of this approach, resistance to PARP inhibition has been observed in majority of patients with advanced cancer. Novel ways
of interrogating PARP resistance are necessary to further elucidate the mechanisms of drug resistance and help identify ways to overcome it.
To probe potential mechanisms of resistance we applied data-independent acquisition (DIA) mass spectrometry for unbiased global protein quantification in patient derived xenografts with demonstrated resistance to PARP inhibitors.
Figure 1: PDX Mouse Models Serve as Tools to Screen Anticancer Drugs and Study Drug Resistance (A) A biobank of PDX models was previously generated and characterized (Bruna et al. Cell 2016) to preserve the intra-tumor heterogeneity and server as a robust platform for pre-clinical pharmacological studies. (B/C) Representative PDX models, both carriers of the PARP susceptibility “BRCAness” molecular characteristics showed differential drug response.
Figure 3: Metabolic, Structural and Neuronal/Embryonic Proteins Identified as Hallmarks of PARP Resistance (A) Volcano plot of 448 human proteins significantly changed between PARP drug response groups. (B) Proteins members of metabolic pathways (ENO1, FASN, LDHB, TBC1D4) structural proteins (VIM, EPPK1, KRT8/5/18/6A) and neuronal/embryonic pathways (DNMT1, UCHL1, NES) were significantly disregulated between the groups.
Figure 2: HRM-MS Discovery Proteomics Quantified 8,500 Proteins from Human and Mouse Origin (A) On average 68.3% of quantified proteins were of human origin and 22.1% of murine origin with comparable ratios across samples. (B) Unsupervised hierarchical clustering of the evaluated models (6 PARP Resistant and 11 PARP Sensitive PDX models) revealed strong heterogeneity based on differential protein abundance.
Figure 4: Functional Analysis Revealed Multiple Dysregulated Interaction Networks (A) DNA damage repair, immune response and non-homologous end joining pathways among the enriched interaction networks. Red color designates higher expression in PARP resistance. (B) Interactome of the DNA damage responses revealed disregulation of CHEK2, TP53, TPS53BP1, RAD50 and H2AFX which were up-regulated in samples resistant to PARP therapy.
Kristina Beeler, PhDHead of Business Development
• PDX Mouse Models Serve as Tools to Screen Anticancer Drugs and Study Drug Resistance
• Hyper Reaction Monitoring Mass
Spectrometry (HRMTM-MS) Discovery Proteomics Quantified 8,500 Proteins from Human and Mouse Origin
• Metabolic, Structural and Neuronal/Embryonic Proteins Identified as Hallmarks of PARP Resistance
To our knowledge, this is the first report aiming at interrogating PARP inhibitor resistance with unbiased quantitative proteomics.
The proteomics data confirmed prior observations of resistance mechanisms to PARP and elucidated potential novel mechanisms involving modulation of the immune response in resistant tumors.
Days of treatment
Sensitive Model
Days of treatment
Resistant Model
Breast Cancer Patients
Patient Derived Xenograft (PDX)
Model
Characteriza-tion of Tumor Heterogeneity
Response to Therapy
Mechanisms of Drug Resistance
Fresh-frozen Tumor Tissue from PDX Mouse Models
Protein Extraction and Trypsin Digestion
Global Proteome Profiling with HRMTM-MS
A B C
A B
-5 0 50
20
40
60
Log2 Fold Change
-log 1
0(Q
-val
ue)
ENO1VIM
VCAN SEC16AUCHL1 AGL
SELENBP1EML4
AKAP1
DNA Damage Response
Immune Response Non-Homo
logous End Joining
TP53
TP53BP1FANCI
ATM
FANCD25’746 807 1’924
Human
MouseHuman Proteome
Mouse Proteome
Neuronal/Embryonic ProteinsStructural ProteinsMetabolic Pathways
Red: Up in ResistanceBlue: Up in Sensitivity
Resistant Models Sensitive Models
A B
A B
68%
22%
10%
Human MouseShared
Red: Up in ResistanceBlue: Up in Sensitivity
FFPE Tissue(2 x 5 µm slices)
-40
-20
0
20
40
0-20 20
standardized component 1
stan
dard
ized
com
pone
nt 2
Non-RespondersResponders
INTRODUCTION
RESULTS
CONCLUSIONS
CHARACTERIZATION OF A DATA INDEPENDENT ACQUISITION MASS SPECTROMETRY-BASED WORKFLOW IN PLASMA
Kamil Sklodowski1, Nicholas Dupuis1, Linda Sensbach1, Sebastian Müller1, Lukas Reiter1
1) Biognosys AG, Wagistrasse 21, 8952 Schlieren (Zurich), Switzerland
Measurement of circulating biomarkers in cancer has proven utility for early detection, differential diagnosis, and predicting pre-treatment response to therapy. Recently, circulating proteomic biomarkers have received additional attention due to the heterogeneous responses to immunotherapies. To develop a greater understanding of the circulating plasma proteome we have optimized a depleted plasma proteomic workflow, based on label-free data-
independent acquisition mass spectrometry (DIA-MS), and applied it to plasma from subjects with late stage NSCLC. This unbiased approach identified proteins of potential diagnostic value for lung cancer.
Figure 1: Outline of study design and rational (A) and details of sample set used (B)
Figure 2: Overview of Assay Workflow (A) and Depletion Optimization (B) (A) Use of commercial cartridges (MARS 14) allows for a standardized and reproducible sample depletion. (B) Subsequent DIA acquisition of samples after depletion allowed for quantification of 1,304 proteins (reference library had 1,827 unique spectra). An improvement of >3 times more proteins quantified than without depletion.
Figure 4: Significantly disregulated proteins (A) and enrichment of biological processes (B)(A) Univariate statistical testing identified 162 dis-regulated proteins (125 up-regulated and 37 down-regulated; q-value > 0.05, log2 fold change > 0.58). (B) Analysis of biological processes of all disregulated proteins showed highest enrichment in acute phase reactions followed by other immune responses (including complement pathway).
Figure 3: Multivariate Analysis of Proteomic Signature in Normal and NSCLC SubjectsMultivariate analysis based on Partial Least Squares Discriminant Analysis (PLS-DA) was able to separate both subject groups. Large variability within disease profiles is observed. Selected proteins from top 25 with biggest effect on the separation are depicted including potential new biomarker candidates like F13A11 involved in macrophage activation, and known targets with immunomodulatory function such as S100A92.
Kamil Sklodowski, PhDClinical Services Manager
• 25 most variable proteins appeared to be linked to host immune response to tumor.
• Univariate tests revealed 162 significantly dis-regulated proteins, while functional analysis showed enrichment in acute phase response.
• ROC analysis of top candidates showed diagnostic value of proteins linked to acute phase response (CRP, C9, SAA1/2 and HPT), immunomodulation (S100A8/9), signaling and metastasis (LRG1), and nutrition (TTR).
Hyper Reaction Monitoring Mass Spectrometry (HRMTM-MS) provided unbiased characterization of plasma proteome:• No depletion >410 proteins, with depletion
>1300 proteins quantified across all samples.Multiple potential biomarkers are identified to be dis-regulated in NSCLC vs Normal subjects:• Multivariate analysis separated diagnostic
subgroups based on proteomic signature.
Figure 5: Receiver Operating Characteristic (ROC) curve (based on dmin between 0 and 1) Potential diagnostic power was evaluated for some of top dis-regulated proteins. Among these acute phase response proteins SAA1/26,9, C93, CRP4,7 and HPT4 have been reported as potential prognostic biomarkers for NSCLC. In case of C93 and SAA1/29 in vitro studies showed their role in macrophage regulation. Calcium binding proteins S100A8/92, known targets with immunomodulatory effects in NSCLC subjects, as
A B A B
A B
Sample Collection and Preparation
Depletion via MARS14 Cartridge
Digestion Global proteome profiling with HRMTM-MS Depleted Plasma
Non-depleted Plasma
NSCLCNormal
Normal
NSCLC
well as LRG15 which is highly expressed in tumor derived exosomes and linked to metastasis, together with TTR8 (not shown), a nutritional biomarker were also found to have a good diagnostic value.
References: 1.Kubach et al., Blood 2007, 110:1550-1558 2.Shen et al., Cancer Immunol Res. 2015, 3(2):136-148 3.Li et al., Cell Death Discov. 2018, 4:63 4.Calabrese et al., J. Clin. Med. 2019, 8:414 5.Alipoor et al., Front. Immunol. 2018, 9:819 6.Liu et al., Biomed Res. Int. 2018, 9815806 7.Walker et al., EBioMedicine, 2015, 2:841–850 8.Shimura et al., Mol. and Clin Onc. 2019, 10:597-604 9.Dakubo et al., 1st ed., Springer, 2017
INTRODUCTION
RESULTS
CONCLUSIONS
PROTEOMIC PROFILING IDENTIFIES PROTEINS ASSOCIATED WITH THERAPEUTIC RESPONSE TO PD-1 IMMUNOTHERAPY
Kristina Beeler1, Nicholas Dupuis1, Jakob Vowinckel1, Domenico Mallardo2, Mariaelena Capone2, Madonna Gabriele2, Antonio Sorrentino2, Vito Vanella2, DanielHeinzmann1, Paolo Ascierto2
1) Biognosys AG, Schlieren (Zurich), Switzerland 2) Istituto Nazionale Tumori IRCCS Fondazione G. Pascale, Naples, Italy
Immune checkpoint inhibitors (ICI) have greatly improved the treatment options for patients with advanced stage melanoma, with improved clinicalresponses and overall survival compared to standard systemic therapies.
However, a large percentage of melanoma patients do not respond to ICIs, highlighting the need for a greater understanding of the tumor environment and host immune response.
Here, we apply unbiased discovery proteomics, based on label-free data-independent acquisition (DIA) mass spectrometry, to deeply characterize global tumor proteomes to identify proteins and pathways that are associated with pretreatmentresponse to anti-PD-1 immunotherapy.
Figure 1: Discovery Proteomics Workflow Based on DIA Mass Spectrometry Quantified > 7,500 Proteins in FFPE Tissue. Proteins were extracted from FFPE samples and prepared for mass spectrometry to generate tryptic peptides. Samples were analyzed on high resolution mass spectrometry instruments and DIA data was extracted using SpectronautTM. Signals were extracted and proteins were quantified.
Figure 2: Global Proteome Analysis Reveals Immune System and Metabolic Processes (A) Partial Least Squares Discriminant Analysis (PLS-DA) identified a subset of proteins driving the difference between responders and non-responders. (B) Box plots of thetop 25 proteins highlight proteins both up and down regulated in the two groups. (C) The top 25 proteins form component 1 are sufficient to reconstruct the responder subgroups.
Figure 3: PLEKHA5 a Regulator of Brain Metastases Associate with Poor Response to PD-1 (A) PLEKHA5 was significantly up-regulated in the non-responder group. Prior studies by Jilaveanu et. al. (Clin Cancer Res; 21(9); 1978–80) had identified that overexpression of PLEKHA5 was associated with cerebrotropic tumors. (B) Brain metastasis free survival for a large cohort of melanoma subjects was stratified based on PLEKHA5 expression .
Kristina Beeler, PhDHead of Business Development
• A pathway level analysis showed increased metabolic processes associated with clinical response to ICI
• PLEKHA5 was strongly associated with non-responder status
• PLEKHA5 expression correlated with brain metastasis
• Global proteomic analysis of FFPE specimens provides deep and unbiased quantification of tumor proteomes
• A set of 25 protein candidates were identified as a proteomic signature associated with response to PD-1 immunotherapy treatment in this initial discovery cohort
A B
Pre-Treatment Biopsies from Melanoma Patients
Protein Extraction and Trypsin Digestion
Global Proteome Profiling with HRMTM-MS*
FFPE Tissue(2x 5 um slices)
A
Proteins Peptides Peptides/Protein
HRM Quantification(Average) 6’627 57’840 8.7
HRM Quantification(Total) 7’590 107’940 14.2
Time (Days)
Overall Survival
Perc
ent (
%)
600 80040020000
Time (Days)
Progression-Free Survival
Perc
ent (
%)
6004002000
100
50
100
50
0
RespondersNon-Responders
Responders
Non-RespondersTime (Days)
Overall Survival
Perc
ent (
%)
600 80040020000
Time (Days)
Progression-Free Survival
Perc
ent (
%)
6004002000
100
50
100
50
0
RespondersNon-Responders
Responders
Non-Responders
Jila
vean
u et
al.
2015
PLEKHA5
Non-Responders Responders Non-Responders Responders
RHOT1SHMT2HSPE1DARS2C1QBPACAD10TUFMTRAP1MRPL20UQCC1KRT9TDRD7GM2ANANSCTSLCTSZPAG1PLEKHA5PLXNC1RABL6TRM32CDKN2AIPNLSNX8HABP4SYTL2
GROUP
PLEKHA5 Expression
Blood-Brain-Barrier Permeability
Brain Metastasis
Quantifiable Proteome in FFPE Sections
C B
Melanoma Patients Treated with First-line PD-1 Targeting Antibodies
Treated Patients Separated into Two Clinically Distinct Groups
* Hyper Reaction Monitoring Mass Spectrometry
INTRODUCTION
RESULTS
CONCLUSIONS
Jakob Vowinckel1, Karel Novy1, Thomas Corwin2, Jonathan Woodsmith2, Tobias Treiber1, Roland Bruderer1, Lukas Reiter1, Eike-Christin von Leitner2, Oliver Rinner1, Claudia Escher1
1) Biognosys AG, Wagistrasse 21, 8952 Schlieren, Switzerland 2) Indivumed GmbH, Falkenried 88, 20251 Hamburg, Germany
Precision oncology requires a detailed molecular understanding of tumor biology. Phenotype and underlying cellular functions are best characterized by the study of the proteome. However, MS-based proteome profiling is underrepresented in precision medicine compared to DNA/RNA sequencing techniques. Limitations in instrument stability, reproducibility, sample throughput and data analysis have prevented large-scale proteome characterization experiments. Recent developments in data-independent acquisition (DIA) LC-MS*/MS
and robust chromatographic separation now present the opportunity to make proteomics available to routine analysis. Here we present a workflow that is capable of routine profiling 850 whole proteome (WP) or 650 phospho-proteome (PP) tumor samples per month with an average depth of 6,000 proteins (WP) or 30,000 phospho-peptides (PP), respectively. The workflow was applied across several indications as depicted on the right. NSCLC is highlighted in this poster.
Figure 1: IndivuType Sample Collection and Processing Workflow (A) Patient bio-specimens and matching clinical data were collected through Indivumed’s global clinical network in a standardized manner. (B) Subset of the clinical attributes for the cohort of ~800 non-small cell lung cancer (NSCLC) patients. 1 person represents 50 patients. (C) Indivumed follows SOP-driven standardized tissue collection approach to minimize molecular alterations resulting from post-surgery tissue collection and preservation processes that allows for an accurate representation of a patient’s tumor biology. Representative immunohistochemistry for pERK 1/2 from one patient taken at three timepoints. pERK 1/2 expression levels increased from 10 minutes to 60 minutes post-surgery demonstrating early molecular changes. (D) Both whole-proteome and phospho-proteome profiling of tumor tissue and adjacent normal tissue from each NSCLC patient was performed using state-of-the-art liquid chromatography-tandem mass spectrometry in DIA mode.
Figure 2: Protein and Phospho-peptide Profiling over 1,755 Samples (A) Hierarchical clustering and (B) principal component analysis of 7,346 protein intensity values reveals co-clustering according to tissue type in 1,755 lung samples. (C) Similar pattern observed in principal component analysis of phospho-peptides. (D) Three known markers of lung cancer show robust change between normal and tumor tissue. EGFR phosphorylation status on three known C-terminal sites (E) are consistently elevated in tumor compared to normal tissue.
Figure 3: NSCLC tumor biology. EGFR signaling pathway is well represented, with 92 out of 120 KEGG annotated proteins quantified in NSCLC tumor samples with protein and/or phospho-peptide information.
Jakob Vowinckel, PhDSenior Scientific Project Manager
• An optimized, semi-automated workflow enables high throughput deep proteome and phospho-proteome profiling of matching tumor and normal tissue samples from Indivumed’s high quality collection of fresh-frozen biospecimens.
• Rigorous quality control during sample collection, sample processing,
data acquisition and analysis allows reproducible generation of data sets consisting of thousands of samples.
• On average, 5,903 protein groups and 28,819 phosphopeptides were quantified in each lung tissue sample within the NSCLC patient cohort, with up to 7,346 protein groups quantified in total.
• Observed protein expression and phosphorylation status of known substrates are in accordance with established knowledge and provide valuable insights to previously unknown markers relevant for tumor biology.
Principal component 1
Prin
cipa
l com
pone
nt 2
Global Clinical Network for Sample Collection Clinical Parameters of NSCLC Patient Cohort
Preserving Molecular Composition andMinimizing Tissue Data Variability
Proteomic Sample Analysis by DIA LC-MS*/MS
A B
CD
A HIGH-THROUGHPUT PLATFORM FOR PROTEOME AND PHOSPHO-PROTEOME PROFILING OF TUMOR TISSUES
B
Principal component 2
Prin
cipa
l com
pone
nt 3
log2
pro
tein
inte
nsity
log2
pep
tide
inte
nsity
EGFR (S1064)
EGFR (Y1110)
n.d.
EGFR (Y1197)
EDC
A
© 2020 by Indivumed GmbH and Biognosys AG. All rights reserved. Released for viewing by AACR Virtual Annual Meeting participants only. Any unauthorized copying, reproduction and/or distribution in whole or in part is prohibited.
1,755 NSCLC Samples
* Liquid Chromatography–Mass Spectrometry