Unravelling the proteome of diffuse large B-cell lymphoma
Meeren, van der, Lotte
DOI:
10.33612/diss.92802769
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Meeren, van der, L. (2019). Unravelling the proteome of diffuse large B-cell lymphoma: Differences in cell of origin and HLA loss. Rijksuniversiteit Groningen. https://doi.org/10.33612/diss.92802769
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
DIFFUSE
LARGE
B-CELL
LYMPHOMA
ISBN: 978-94-6332-493-9
Cover design & lay-out: Esther Beekman (www.estherontwerpt.nl) Printed by: GVO drukkers & vormgevers B.V., Ede
© 2019 Lotte-Elisabeth van der Meeren
All rights reserved. No part of this dissertation may be reprinted, reproduced, or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording or any information storage or retrieval system, without prior written permission of the author.
Financial support for the printing of this thesis was kindly provided by
Differences in cell of origin and HLA loss
Proefschrift
ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen
op gezag van de
rector magnificus prof. dr. C. Wijmenga en volgens besluit van het College voor Promoties.
De openbare verdediging zal plaatsvinden op woensdag 11 september 2019 om 11.00 uur
door
Lotte-Elisabeth van der Meeren geboren op 23 juni 1985
Dr. L. Visser
Beoordelingscommissie Prof. dr. H. Hollema Prof. dr. S.T. Pals Prof. dr. E. Vellenga
Chapter 3
Chapter 4
Chapter 5
Chapter 6
Chapter 7
Combined loss of HLA I and HLA II expression is more
common in the non-GCB type of diffuse large B-cell lymphoma Pr oteomics analysis to identify differences between HLA class I and HLA class II positive and negative diffuse large B-cell lymphoma patient samples using super-SILAC
Global secretome and membrane protein analysis of follicular and diffuse large B-cell lymphoma
Summary, discussion & future perspectives Nederlandse samenvatting
52
62
82
102
122
1
Introduction and
scope of thesis
10 11
PATHOLOGY AND CLASSIFICATION
Non-Hodgkin lymphomas account for 40-50% of all hematopoietic malignancies and within the group of non-Hodgkin lymphomas approximately 90% are of B-lymphocyte lineage. The most common type of non-Hodgkin lymphoma (30-40% in western countries) is diffuse large B-cell lymphoma (DLBCL). DLBCL can present de novo or transform from a less aggressive lymphoma such as follicular lymphoma, marginal zone lymphoma or nodular lymphocyte predominant Hodgkin lymphoma. DLBCL can be further divided based on different aspects such as clinical presentation and behaviour, histology and/or presence of Epstein Barr Virus (EBV). Examples of distinct subtypes are primary DLBCL of the central nervous system, primary cutaneous DLBCL leg type, T-cell/histiocytic rich B-cell lymphoma and EBV positive DLBCL of the elderly.
1 However, these subtypes are all relatively rare, because the majority of the patients
do not fulfil the criteria of these specific subgroups. The largest unclassified subgroup is defined as DLBCL ‘not otherwise specified’ (DLBCL NOS). 1
DLBCL is histologically defined by a diffuse proliferation of large B-cells that have completely or partially altered the lymph node architecture, the pre-existent architecture, and in case of lymph node involvement, often with infiltration of
perinodal tissue. 2,3 The large tumour cells usually present with a nuclear size at least
twice the size of a normal lymphocyte with vesicular, prominent nuclei and pale or basophilic cytoplasm. Mitotic figures or apoptotic cells may be present. Tumour cells usually express the pan B-cell markers CD19, CD20, CD79a and PAX5, and show a variable expression of markers such as the germinal centre B-cell associated BCL6 (60-80%) and CD10 (40%) and the more post germinal centre marker IRF4/MUM1 (40%). BCL6 and CD10 are both involved in the induction and maintenance of germinal centres. 2,4 The heterogeneity in morphology, phenotype, clinical
presentation and behaviour of DLBCL NOS suggested that this is a heterogeneous group that should be further classified in distinct subtypes. In 2000, DLBCL NOS was divided into two distinct entities based on their gene expression profiles: germinal centre B-cell like (GCB) DLBCL and activated B-cell like (ABC or non-GCB) DLBCL. These molecular subtypes have different prognostic outcomes; the non-GCB DLBCL subtype is associated with a poor prognosis. 5,6
GCB DLBCL arises from B-cells in the germinal centre stage of maturation. In 25%-35% of all DLBCL a translocation of BCL6 is present; however, this translocation is not specific for GCB DLBCL. 4,7 Other alterations include mutations in the
10 11
PI3K/mTOR/AKT pathway (GNA11, SGK1 and PTEN), epigenetic regulation (EZH2 and MEF2B), immune regulation (TNFRSF14) and apoptosis (BCL2). 8 The t(14;18)
is more frequently found in GCB-DLBCL (30%-45%) compared to non-GCB DLBCL group (4,5%) and results in overexpression of the anti-apoptotic BCL2 gene. 9,10
Approximately 20% of GCB DLBCL cases have an EZH2 mutation. 11,12 EZH2 is part
of the polycomb protein subunit PRC2. 11,12 This results in trimethylation of histone
H3K27 (H3K27me3), which is associated with a transcriptional repressive state of the
locus. 13 MYC translocations are detected in 6%-14% of DLBCL patients and are more
common in GCB DLBCL (10%) compared to non-GCB DLBCL (5%). 13,14 In relapsed
DLBCL NOS MYC translocations are present in an even higher proportion of the patients, i.e. 20%. 15,16 Activation of MYC can deregulate proliferation and alters the
metabolic state of the tumour cells, leading to malignant transformation of cells. Moreover, MYC regulates a broad range of other genes including the anti-apoptotic BCL2 gene. 4,7,13
Non-GCB DLBCL arises from post-germinal B-cells that are blocked during plasmocytic differentiation. 17,18 A characteristic finding in non-GCB DLBCL is
activation of the NF-ĸB signalling pathway that controls cell survival and inhibits apoptosis. Genes commonly mutated in non-GCB DLBCL include genes of the BCR/NF-ĸB signalling pathway (CD79a, CD79b and MYD88). 8 MYD88 mutations
are present in over 30% of non-GCB DLBCL and lead to activation of the NF-ĸB signalling pathway. 16 This subgroup is further characterized by gain of 3q (26%)
and gain of the BCL2 gene locus at 18q21-q22 (18%). 14,19 A second way to activate
the NF-ĸB signalling pathway is by triggering of the CARD11-BCL10-MALT (CBM) signalling complex. 10,20,21 This CBM complex can be activated by mutations in
CARD11 (10%) and chronic activation of the B-cell receptor due to mutations in CD79a or CD79b. 18,19
CLINICAL PRESENTATION, TREATMENT
AND PROGNOSTIC FACTORS
DLBCL can occur at all ages, but is most frequent in patients older than 60 years and slightly more common in males. 1,22 Up to 40% of patients present initially with
extra nodal disease, most commonly the gastrointestinal tract. Other less frequently involved sites are testis, bone, skin, spleen, liver and kidney. Some patients present
12 13
with B-symptoms such as weight loss and fever due to the release of cytokines by the tumour cells. The Ann Arbor system is used to stage the patients (stage I with localized disease to stage IV with widely disseminated disease). The international prognostic index (IPI) is used to classify patients based on clinical factors such as age, stage of the disease, serum LDH level, performance status and location of tumour (more than one extra nodal site). The IPI score is used to identify prognostic subgroups and predict outcome. 23,24 The currently used standard chemotherapy is
R-CHOP (Rituximab - Cyclophosphamide, Doxorubicin, Vincristine and Prednisolone).
24,25 The overall survival of patient treated with R-CHOP in the prognostic favourable
group (IPI 0-1) is 97%, whereas the overall survival in the prognostic poor group (IPI 4-5) is only 55%. 23,26 Patients with therapy failure usually have partial remission of the
disease or present with early relapse. 27
MICROENVIRONMENT AND
MODULATION OF TUMOUR
DEVELOPMENT BY THE IMMUNE
SYSTEM
The pathogenesis of DLBCL NOS is a multistep process that involves genetic alterations, immune status of the patient and tumour microenvironment. The microenvironment includes stromal cells, intra-tumour vasculature, macrophages, follicular dendritic cells, follicular reticular cells and various subsets of T-cells. 18,28,29
This is illustrated by the inclusion of T-cell/histiocytic rich B-cell lymphoma as a separate entity in the WHO classification and the numerous publications on the influence of the microenvironment on global gene expression. 18,30-33
The concept of tumour cell interaction with the microenvironment exists for over 100 years. In 1893 William Coley was presumably the first who published the
hypothesis that the immune system can modulate tumour development. He injected toxins into a patient with cancer to stimulate the immune system to establish an anti-tumour immune response. 34 In 1909 Paul Erhlich suggested that immune responses
could be modulated or suppressed by the tumour microenvironment. 35,36 Nowadays
it is commonly accepted that the microenvironment plays an important role in the malignant transformation and survival of tumour cells. The microenvironment can
12 13
stimulate tumour cell survival or promote tumour regression, depending on the functionality and activity of the infiltrating cells.
A significant correlation of the microenvironment with the behaviour of non-Hodgkin lymphomas has been reported for some years. 28 Several studies suggest that the
microenvironment creates a niche that can regulate cancer cell proliferation. Alizadeh et al. and Rosenwald et al. identified signatures linked to immune response and genes involved in remodelling the extracellular matrix and connective tissue growth factors. 17,28,29 Lenz et al. reported a multivariate model with three gene-expression
signatures. 18 They compared DLBCL NOS treated with CHOP and R-CHOP and
identified two stromal signatures; stromal-1 and stromal-2 that were especially important for the microenvironment and variably present in both GCB DLBCL and non-GCB DLBCL. The stromal-1 signatures identified DLBCL cases with extensive extracellular matrix and infiltration by cells of the monocytic lineage and are related to a favourable prognosis. The stromal-2 signature includes endothelial cell markers and various genes necessary for angiogenesis and is associated with an adverse outcome. 37 The significance of follicular dendritic cell networks present in DLBCL
was investigated by Van Imhoff et al. 38 They proposed that CD21+ follicular dendritic
cell networks might indicate an interaction of tumour cells with the surrounding microenvironment or an origin from a follicular or other indolent lymphoma. Presence of this network was associated with an increased relapse rate. 38 Linderoth et al.
compared de novo DLBCL in complete remission versus progressive disease during primary treatment and identified genes associated with treatment resistance. 39 The
number of tumour infiltrating lymphocytes, macrophages, and reactive cells in the surrounding tumour microenvironment was significantly higher in the DLBCL cohort with complete remission implying that an active microenvironment influences the treatment outcome. The underlying mechanisms and relevant interactions are still not completely known.
CYTOKINES & CHEMOKINES
Cytokines are produced by B-cells, T-cells and macrophages and are involved in the interaction with tumour and microenvironment. Cytokines are part of various signalling pathways and regulate proliferation and differentiation. An active microenvironment with expression of proteins involved in cytokine signalling,
14 15
seen in DLBCL with an aggressive phenotype and a poor prognosis. 39,40 Cytokines
and chemokines can be found in serum and can be measured relatively easily. Therefore, serum analysis could be an elegant option to predict prognosis or monitor therapy effect. Riby et al. analysed sera from DLBCL patients using liquid chromatography – mass spectrometry (LC-MS) to identify cytokines significantly elevated in DLBCL patients to explore the possibility to develop non-invasive
biomarker panels. 41 For example, CXCL10 and CCL4 are chemokines found in serum
and promote tumour growth and invasion of tumour by attracting inflammatory cells in the microenvironment. 42,43 High levels of serum CXCL10 correlated with an
aggressive behaviour in DLBCL. 40 High levels of serum MIF were associated with
tumour cell survival by preventing an adequate cytotoxic T-cell reaction. 44
HLA AND THE IMMUNE SYSTEM IN
DLBCL
The Human Leukocyte Antigen (HLA) system regulates the immune response and loss of HLA expression in the tumour cells leads to an altered host immunity and impairs effective anti-tumour responses. B-cells express both HLA class I and HLA class II. However, during malignant transformation or evolution of B-cell lymphomas, expression of HLA class I and/or HLA class II may be lost. 1 Loss of HLA class I and
HLA class II expression has been described in various subtypes of DLBCL and is caused by several mechanisms. Combined loss of HLA class I and HLA class II is usually caused by small deletions at chromosome 6. 33,45 Loss of HLA class I is mainly
caused by mutations and deletions in ß2-microglobulin 46 and loss of HLA class II
is caused by loss of CIITA expression, which is downregulated due to epigenetic modifications. 47,48 In DLBCL individual or combined loss of HLA I and HLA II is most
common in non-GCB DLBCL. 49 Several studies have shown that loss of HLA class I
and HLA class II correlates with poor prognosis and overall survival, both in CHOP and R-CHOP treated patients. 50,51
14 15
OVERVIEW OF THE HISTORY OF
PROTEOMICS
The first description of proteins in the literature was in 1789 by Antoine Fourcroy, who described proteins such as albumin, fibrin, gelatin and gluten. 52 In 1837, the
Dutch chemist Gerrit J. Mulder determined the elemental composition of several proteins. Following this, Jacob Berzelius proposed in 1838 the name protein from the Greek prwteioz meaning “standing in front”, “in the lead”. 52 The term proteome
is a combination of two words, protein and genome and was introduced at the 1st Siena meeting in 1994 by Marc Wilkins. 53,54 Proteomics refers to large-scale
analysis of the proteome, attempting to better characterize protein function and identify interactions with other proteins. Qualitative proteomics approaches result in identification of proteins in a given sample, whereas quantitative proteomics has an additional dimension including both identification and quantification. Based on specific research goals several proteomics approaches and strategies can be applied. Targeted proteomics is suitable for known targets. This technique has a high sensitivity and is used for analysing small numbers of proteins, usually less than 100. Discovery proteomics or shotgun proteomics refers to a global intensive analysis based on signal intensity to discover as many proteins as possible. 55 The currently
available techniques give a reliable high throughput outcome resulting in a huge amount of data. Therefore, integration of this information needs interdisciplinary research. To allow for a comprehensive analysis it is crucial to integrate bioinformatics and mathematics and combine proteomics data with existing techniques such as gene expression profiling. 56,57
SUPER-SILAC
Stable isotope labelling with amino acids in cell culture (SILAC) is a quantitative bottom-up proteomics method with metabolically labelled cells to identify proteins and quantify their expression levels. 58,59 Cells are metabolically labelled by growing
them in medium supplemented with labelled ‘heavy’ lysine and arginine. With every passage labelled amino acids are incorporated into newly synthesized proteins, to allow maximum incorporation of labelled amino acids into all proteins six to eight passages are required. 59 Therefore it can only be used for cell lines, which can be
16 17
proteins in a 1:1 protein ratio, it is possible to directly compare protein levels; labelled with ‘heavy’ 13C
6 L-Lysine-2HCl and 13C615N4 L-Arginine-HCl or ‘light’ normal
lysine and arginine. A disadvantage of SILAC is that it is not possible to directly quantify differences between samples that cannot be cultured. With the introduction of super-SILAC 60 it has become possible to quantify and characterize proteins in
samples without the need to culture the cells. A mixture of ‘heavy’ labelled proteins of cell lines is made and used as a normalization control. Each sample is mixed with this control allowing direct comparison of independent samples. To allow optimal detection of proteins a mix of multiple cell lines should be used to create a super-SILAC mix with sufficient complexity. Protein identification and heavy/light ratios are measured by LC-MS. The differences in protein expression levels between samples are indicated by the ratio of ‘heavy’ isotopically labelled and ‘light’ non-labelled peak intensities in the mass spectrum for a specific protein.
ANALYSIS OF DLBCL USING
SUPER-SILAC
Three large proteomics-based studies have been published to identify differentially expressed proteins in DLBCL cell lines and patient samples. Deeb et al. characterized DLBCL cell lines with super-SILAC to differentiate between GCB DLBCL and non-GCB DLBCL. 61 They analysed 10 DLBCL cell lines (OCI-LY3, RIVA, U2932, TMD8,
HBL-1, BJAB, SU-DHL-4, SU-DHL-6, DB and HT) with LC-MS and identified 7756 proteins. They established a signature of 55 proteins that segregated the GCB DLBCL and non-GCB cell lines. This included known proteins such as IRF4 and CD44 involved in the NF-κB pathway for the non-GCB subtype. In their protein signature two upstream regulatory proteins of the CBM signalling complex were present; CARD11 and MALT. PTP1B, a tyrosine phosphatase implicated in the JAK/STAT pathway showed a higher expression in non-GCB DLBCL cell lines. Two proteins were differently expressed in GCB DLBCL compared to non-GCB DLBCL cell lines; CD81 and SPI1. SPI1 plays a role in maintenance of germinal centre B-cells by repressing plasma cell transcription factors. 61 In a second study, Deeb et al. used super-SILAC to characterize
formalin-fixed paraffin-embedded tissues of 20 patients with GCB DLBCL or non-GCB DLBCL.
62 This revealed 343 differentially expressed proteins. They could classify 15 out
of the 20 cases based on their proteome into the correct subtype. Ruetschi et al. have analysed frozen tissues of five relapsed and five long-term progression-free
16 17
DLBCL patients using super-SILAC. Five cell lines were selected for the reference super-SILAC mix; four DLBCL cell lines (SC-1, WSU-NHL, SU-DHL-5, and SU-DHL-8) and one Burkitt’s lymphoma cell line (DG-75). 63 Of the 3588 proteins, 87 were
differentially expressed between the two patient groups (p<0.05). They validated differential expression for 3 proteins by western blot on the same patients. 63 These
three proteins identified in their signature were involved in organisation of the actin cytoskeleton: annexin VI, cap1 and moesin.
AIM
The overall aim of this thesis is to investigate the proteome of DLBCL to identify proteins that can discriminate between GCB DLBCL and non-GCB DLBCL, are differentially expressed in HLA positive and negative DLBCL cases and play a role in the cross talk between tumour cells and the microenvironment.
OUTLINE
This thesis describes the results of large-scale proteomics analyses of DLBCL cell lines and primary DLBCL patient samples. Chapter 1 is a general introduction of diffuse large B-cell lymphoma with an outline of the characteristic findings of DLBCL. In chapter 2 we applied super-SILAC to analyse thirteen DLBCL patient samples with the aim to identify differences between GCB and non-GCB DLBCL patient samples. Results were validated by western blot and IHC on the same cases and by immunohistochemistry on an independent patient cohort. In chapter 3 we studied loss of HLA class I and HLA class II expression in GCB DLBCL and non-GCB DLBCL cases as assessed by IHC. In chapter 4 we re-analysed our super-SILAC data of chapter 2, to identify proteins differentially expressed between cases with and without combined HLA class I and HLA class II loss. Results were validated by immunohistochemistry in the same and an independent patient cohort. In chapter 5 we investigated the secretome of various aggressive B-cell lymphoma cell lines including cell lines derived from diffuse large B-cell lymphoma. Finally, chapter 6 presents a summary and overall discussion on the studies performed in this thesis and future perspectives with suggestions for follow-up studies.
18 19
REFERENCES
1. Swerdlow SH, Campo EB, Harris NL, Jaffe ES, Pileri SA, Stein H, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. 4 ed. Vol. 2. 2017.
2. Vinay, K, Abul, A, Fausto, Jon N, et al. Robbins and Cotran Pathologic Basis of Disease.
3. Hoffbrand AV, Pettit JE, Vyas P. Clinical Hematology.
4. Carbone A, Gloghini A, Aiello A, Testi A, Cabras A. B-cell lymphomas with features intermediate between distinct pathologic entities. From pathogenesis to pathology. Human Pathology. 2010 May;41(5):621–31.
5. Sjö LD, Poulsen CB, Hansen M, Møller MB, Ralfkiaer E. Profiling of diffuse large B-cell lymphoma by immunohistochemistry: identification of prognostic subgroups. Eur J Haematol. 2007 Dec;79(6):501–7.
6. Sehn LH. Early detection of patients with poor risk diffuse large B-cell lymphoma. Leuk Lymphoma. 2009 Nov;50(11):1744–7.
7. Chan WJC. Pathogenesis of diffuse large B cell lymphoma. Int J Hematol. 2010 Jun 29;92(2):219–30.
8. Li S, Young KH, Medeiros LJ. Diffuse large B-cell lymphoma. Pathology. 2018 Jan 1;50(1):74–87.
9. Bea S. Diffuse large B-cell lymphoma subgroups have distinct genetic profiles that influence tumor biology and improve gene-expression-based survival prediction. Blood. 2005 Nov 1;106(9):3183–90.
10. Sehn LH, Gascoyne RD. Diffuse large B-cell lymphoma: optimizing outcome in the context of clinical and biologic heterogeneity. Blood. 2015 Jan 1;125(1):22–32. 11. Béguelin W, Popovic R, Teater M, Jiang Y, Bunting KL, Rosen M, et al. EZH2 Is
Required for Germinal Center Formation and Somatic EZH2 Mutations Promote Lymphoid Transformation. Cancer Cell. 2013 May 13;23(5):677–92.
12. Velichutina I, Shaknovich R, Geng H, Johnson NA, Gascoyne RD, Melnick AM, et al. EZH2-mediated epigenetic silencing in germinal center B cells contributes to proliferation and lymphomagenesis. Blood. 2010 Dec 9;116(24):5247–55.
13. Meyer N, Penn LZ. Reflecting on 25 years with MYC. Nat Rev Cancer. 2008 Dec;8(12):976–90.
14. Yoon SO, Jeon YK, Paik JH, Kim WY, Kim YA, Kim JE, et al. MYCtranslocation and an increased copy number predict poor prognosis in adult diffuse large B-cell lymphoma (DLBCL), especially in germinal centre-like B cell (GCB) type. Histopathology. 2008 Aug;53(2):205–17.
18 19
15. Thieblemont C, Briere J. MYC, BCL2, BCL6 in DLBCL: impact for clinics in the future? Blood. 2013 Mar 21;121(12):2165–6.
16. Cuccuini W, Briere J, Mounier N, Voelker HU, Rosenwald A, Sundstrom C, et al. MYC+ diffuse large B-cell lymphoma is not salvaged by classical R-ICE or R-DHAP followed by BEAM plus autologous stem cell transplantation. Blood. 2012 May 17;119(20):4619–24.
17. Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature. 2000 Feb 3;403(6769):503–11.
18. Lenz G, Wright G, Dave SS, Xiao W, Powell J, Zhao H, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med. 2008 Nov 27;359(22):2313–23.
19. Testoni M, Zucca E, Young KH, Bertoni F. Genetic lesions in diffuse large B-cell lymphomas. Annals of Oncology. 2015;26(6):1069–80.
20. Davis RE, Brown KD, Siebenlist U, Staudt LM. Constitutive Nuclear Factor κB Activity Is Required for Survival of Activated B Cell–like Diffuse Large B Cell Lymphoma Cells. J Exp Med. 2001 Dec 17;194(12):1861–74.
21. Baldwin AS. Control of oncogenesis and cancer therapy resistance by the transcription factor NF-kappaB. J Clin Invest. 2001 Feb;107(3):241–6.
22. Lowenberg B, Ossenkoppele GJ, De Witte T, Boogaerts MA. Handboek Hematologie.
23. Sehn LH, Berry B, Chhanabhai M, Fitzgerald C, Gill K, Hoskins P, et al. The revised International Prognostic Index (R-IPI) is a better predictor of outcome than the standard IPI for patients with diffuse large B-cell lymphoma treated with R-CHOP. Blood. 2007 Mar 1;109(5):1857–61.
24. Ziepert M, Hasenclever D, Kuhnt E, Glass B, Schmitz N, Pfreundschuh M, et al. Standard International Prognostic Index Remains a Valid Predictor of Outcome for Patients With Aggressive CD20+ B-Cell Lymphoma in the Rituximab Era. Journal of Clinical Oncology. 2010 May 6;28(14):2373–80.
25. Sehn LH. Introduction of Combined CHOP Plus Rituximab Therapy Dramatically Improved Outcome of Diffuse Large B-Cell Lymphoma in British Columbia. Journal of Clinical Oncology. 2005 Jun 6;23(22):5027–33.
26. Dotan E, Aggarwal C, Therapeutics MSPA, 2010. Impact of rituximab (Rituxan) on the treatment of B-cell non-Hodgkin’s lymphoma. Pharmacy and Therapeutics 2010 Mar;35(3):148-57.
27. Muris J, Meijer C, Vos W, van Krieken J, Jiwa N, Ossenkoppele G, et al. Immunohistochemical profiling based on Bcl-2, CD10 and MUM1 expression improves risk stratification in patients with primary nodal diffuse large B cell lymphoma. J Pathol. 2006;208(5):714–23.
20 21
28. Alizadeh AA, Gentles AJ, Alencar AJ, Liu CL, Kohrt HE, Houot R, et al. Prediction of survival in diffuse large B-cell lymphoma based on the expression of 2 genes reflecting tumor and microenvironment. Blood. 2011 Aug 4;118(5):1350–8.
29. Rosenwald A, Wright G, Chan WC, Connors JM, Campo E, Fisher RI, et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med. 2002 Jun 20;346(25):1937–47.
30. Apoorva F, Loiben AM, Shah SB, Purwada A, Fontan L, Goldstein R, et al. How Biophysical Forces Regulate Human B Cell Lymphomas. CellReports. 2018 Apr 10;23(2):499–511.
31. Shaffer AL, Rosenwald A, Hurt EM, Giltnane JM, Lam LT, Pickeral OK, et al. Signatures of the immune response. Immunity. 2001;15(3):375–85.
32. Rimsza L. Regulating the suppressors: Helpful or harmful? Leuk Lymphoma. 2008 Jan;49(2):179–80.
33. Riemersma SA, Oudejans JJ, Vonk MJ, Dreef EJ, Prins FA, Jansen PM, et al. High numbers of tumour-infiltrating activated cytotoxic T lymphocytes, and frequent loss of HLA class I and II expression, are features of aggressive B cell lymphomas of the brain and testis. J Pathol. 2005;206(3):328–36.
34. Coley WB. The Classic: The Treatment of Malignant Tumors by Repeated Inoculations of Erysipelas: With a Report of Ten Original Cases. Clinical orthopaedics and related research. 1991;262:3–11.
35. Dunn GP, Bruce AT, Ikeda H, Old LJ, Schreiber RD. Cancer immunoediting: from immunosurveillance to tumor escape. Nat Immunol. 2002 Nov;3(11):991–8.
36. Ehrlich P. Ueber den jetzigen stand der karzinomforschung. 1909 ed. Nederlands Tijdschrift voor Geneeskunde; 1909. 18 p.
37. Lenz G, Wright GW, Emre NCT, Kohlhammer H, Dave SS, Davis RE, et al. Molecular subtypes of diffuse large B-cell lymphoma arise by distinct genetic pathways. Proc Natl Acad Sci USA. 2008 Sep 9;105(36):13520–5.
38. van Imhoff GW, Boerma EJG, van der Holt B, Schuuring E, Verdonck LF, Kluin-Nelemans HC, et al. Prognostic Impact of Germinal Center-Associated Proteins and Chromosomal Breakpoints in Poor-Risk Diffuse Large B-Cell Lymphoma. Journal of Clinical Oncology. 2006 Aug 14;24(25):4135–42.
39. Linderoth J, Edén P, Ehinger M, Valcich J, Jerkeman M, Bendahl P-O, et al. Genes associated with the tumour microenvironment are differentially expressed in cured versus primary chemotherapy-refractory diffuse large B-cell lymphoma. Br J Haematol. 2008 May;141(4):423–32.
40. Hong JY, Ryu KJ, Lee JY, Park C, Ko YH, Kim WS, et al. Serum level of CXCL10 is associated with inflammatory prognostic biomarkers in patients with diffuse large B-cell lymphoma. Hematological Oncology. 2016 Dec 12;35(4):480–6.
20 21
41. Riby J, Mobley J, Zhang J, Bracci PM, Skibola CF. Serum protein profiling in diffuse large B-cell lymphoma. Prot Clin Appl. 2016 Nov;10(11):1113-1121.
42. Takahashi K, Sivina M, Hoellenriegel J, Oki Y, Hagemeister FB, Fayad L, et al. CCL3 and CCL4 are biomarkers for B cell receptor pathway activation and prognostic serum markers in diffuse large B cell lymphoma. Br J Haematol. 4 ed. 2015 Sep 11;171(5):726–35.
43. Ansell SM, Maurer MJ, Ziesmer SC, Slager SL, Habermann TM, Link BK, et al. Elevated pretreatment serum levels of interferon-inducible protein-10 (CXCL10) predict disease relapse and prognosis in diffuse large B-cell lymphoma patients. Am J Hematol. 2012 Jun 3;87(9):865–9.
44. Abe R, Peng T, Sailors J, Bucala R, Metz CN. Regulation of the CTL Response by Macrophage Migration Inhibitory Factor. J Immunol. 2001 Jan 15;166(2):747–53. 45. Booman M, Szuhai K, Rosenwald A, Hartmann E, Kluin-Nelemans H, de Jong
D, et al. Genomic alterations and gene expression in primary diffuse large B-cell lymphomas of immune-privileged sites: the importance of apoptosis and immunomodulatory pathways. J Pathol. 2008 Oct;216(2):209–17.
46. Challa-Malladi M, Lieu YK, Califano O, Holmes AB, Bhagat G, Murty VV, et al. Combined Genetic Inactivation of β2-Microglobulin and CD58 Reveals Frequent Escape from Immune Recognition in Diffuse Large B Cell Lymphoma. Cancer Cell. 2011 Dec;20(6):728–40.
47. Cycon KA, Rimsza LM, Murphy SP. Alterations in CIITA constitute a common mechanism accounting for downregulation of MHC class II expression in diffuse large B-cell lymphoma (DLBCL). Experimental Hematology. ISEH - Society for Hematology and Stem Cells; 2009 Feb 1;37(2):184–194.e2.
48. Cycon KA, Mulvaney K, Rimsza LM, Persky D, Murphy SP. Histone deacetylase inhibitors activate CIITA and MHC class II antigen expression in diffuse large B-cell lymphoma. Immunology. 2013 Sep 12;140(2):259–72.
49. van der Meeren LE, Visser L, Diepstra A, Nijland M, van den Berg A, Kluin PM. Combined loss of HLA I and HLA II expression is more common in the non-GCB type of diffuse large B cell lymphoma. Histopathology. 2018;72(5):886–8.
50. Bernd HW, Ziepert M, Thorns C, Klapper W, Wacker HH, Hummel M, et al. Loss of HLA-DR expression and immunoblastic morphology predict adverse outcome in diffuse large B-cell lymphoma - analyses of cases from two prospective randomized clinical trials. Haematologica. 2009 Oct 30;94(11):1569–80.
51. Higashi M, Tokuhira M, Fujino S, Yamashita T, Abe K, Arai E, et al. Loss of HLA-DR expression is related to tumor microenvironment and predicts adverse outcome in diffuse large B-cell lymphoma. Leuk Lymphoma. 2015 Jul 7;57(1):161–6.
22 23
52. Tanford C, Reynolds J. Nature’s Robots - A History of Proteins. Oxford University Press.
53. Williams KL, Gooley AA, Wilkins MR, Packer NH. A Sydney proteome story. Journal of Proteomics. 2014 Jul;107:13–23.
54. Godovac-Zimmermann J. 8th Siena meeting. From genome to proteome: integration and proteome completion. Expert Review of Proteomics. (12):1478– 9450.
55. Domon B, Aebersold R. Options and considerations when selecting a quantitative proteomics strategy. Nat Biotechnol. 2010 Jul 9;28(7):710–21.
56. Sidoli S, Cheng L, Jensen ON. Proteomics in chromatin biology and epigenetics: Elucidation of post-translational modifications of histone proteins by mass spectrometry. Journal of Proteomics. 2012 Jun 27;75(12):3419–33.
57. Nesvizhskii AI. Proteogenomics: concepts, applications and computational strategies. Nature Methods. 2014 Nov 1;11(11):1114–25.
58. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 2002 May;1(5):376–86. 59. Ong S-E, Mann M. A practical recipe for stable isotope labeling by amino acids in
cell culture (SILAC). Nat Protoc. 2007 Jan;1(6):2650–60.
60. Geiger T, Cox J, Ostasiewicz P, Wisniewski JR, Mann M. Super-SILAC mix for quantitative proteomics of human tumor tissue. Nature Methods. 2010 Apr 4;7(5):383–5.
61. Deeb SJ, D’Souza RCJ, Cox J, Schmidt-Supprian M, Mann M. Super-SILAC allows classification of diffuse large B-cell lymphoma subtypes by their protein expression profiles. Molecular & Cellular Proteomics. 2012 Feb 22;11(5):77–89.
62. Deeb SJ, Tyanova S, Hummel M, Schmidt-Supprian M, Cox J, Mann M. Machine Learning-based Classification of Diffuse Large B-cell Lymphoma Patients by Their Protein Expression Profiles. Molecular & Cellular Proteomics. 2015 Nov 1;14(11):2947–60.
63. Rüetschi U, Stenson M, Hasselblom S, Nilsson-Ehle H, Hansson U, Fagman H, et al. SILAC-Based Quantitative Proteomic Analysis of Diffuse Large B-Cell Lymphoma Patients. International Journal of Proteomics. 2015;2015:1–12.
2
A super-SILAC based
proteomics analysis
of
diffuse large
B-cell lymphoma
patient samples to
identify new proteins
that discriminate
GCB and non-GCB
lymphomas
L.E. van der Meeren
1,2, J. Kluiver
1, B. Rutgers
1, Y. Alsagoor
1,
P.M. Kluin
1, A. van den Berg
1and L. Visser
1.
1 Department of Pathology and Medical Biology, University of Groningen,
University Medical Centre Groningen, Groningen, the Netherlands.
2 Department of Pathology, University Medical Centre Utrecht, Utrecht,
the Netherlands.
26 27
ABSTRACT
Diffuse large B-cell lymphoma - not otherwise specified (DLBCL NOS) is a large and heterogeneous subgroup of non-Hodgkin lymphoma. DLBCL can be subdivided into germinal centre B-cell like (GCB) and activated B-cell like (non-GCB) using a gene-expression based or an immunohistochemical approach. In this study we aimed to identify additional proteins that are differentially expressed between GCB and non-GCB DLBCL. We made a reference super-SILAC mix including 8 B-cell lymphoma cell lines and used this mix to quantify protein levels in purified tumour cells of 7 non-GCB DLBCL and 5 non-GCB DLBCL patient samples. Protein identification was performed by LC-MS. We identified 5627 proteins with a four-fold significant difference in
expression between non-GCB and GCB DLBCL for 37 proteins. Five proteins were selected for further validation on 10 of the 12 cases by immunohistochemistry and replication in an independent cohort of 47 DLBCL patients. The validation cohort showed only a trend towards the same differential expression as the proteomics cohort, due to the small size of the cohort. The replication study showed significant and consistent differences for 2/4 proteins: expression of glomulin (GLMN) was higher in GCB type DLBCL, while expression of ribosomal protein L23 (RPL23) were both higher in non-GCB DLBCL. These proteins are functionally linked to important pathways involving MYC, p53 and angiogenesis. We showed increased expression of RPL23 and decreased expression of GLMN in non-GCB compared to GCB DLBCL on purified primary DLBCL patient samples and replicated these results in an
26 27
INTRODUCTION
Gene-expression profiling (GEP) has aided our understanding of the pathogenesis of diffuse large B-cell lymphoma not otherwise specified (DLBCL NOS) by discriminating two distinct entities; germinal centre B-cell like (GCB) DLBCL and activated B-cell like (ABC) or non-GCB DLBCL. 1-3 Non-GCB DLBCL arises from post-germinal B-cells
that are blocked during plasmocytic differentiation. 3 Several
immunohistochemistry-based algorithms have been developed to classify DLBCL into non-GCB and GCB subgroups. 4-7 The Hans algorithm is the most commonly used approach to classify
DLBCL cases in the routine diagnostic setting. 8 So far, most proteomics studies
focused on DLBCL cell lines with some exceptions of studies using DLBCL tissue samples. 9-11 Deeb et al. were the first to characterise DLBCL cell lines with
super-SILAC, a quantitative proteomics approach, to differentiate between non-GCB and GCB DLBCL. 9 They defined a list of 55 proteins to segregate non-GCB and GCB
DLBCL cell lines. We applied super-SILAC on purified tumour cells to analyse the membrane and cytoplasmic compartments in search for novel proteins that can further discriminate between non-GCB and GCB DLBCL subgroups. 12 In addition,
we applied immunohistochemistry to validate our results to determine their use in a clinical setting and replicated our findings in an independent cohort of DLBCL patient samples.
MATERIALS AND METHODS
Selection of cases
We collected viably frozen cell suspensions of 59 DLBCL cases in the tissue bank of the department of Pathology UMCG between 1999 and 2012. All cases were reviewed based on the 2008 WHO classification (table 1). 13 Based on an initial
estimation using H&E staining and immunohistochemistry on paraffin and frozen tissue sections, 31/59 cases had sufficient numbers of tumour cells (approximately >80% tumour) for proteomics analysis. From these cases, sufficient viable single cell suspensions were available for 12 cases. An independent cohort of 47 DLBCL NOS cases (table 2) was selected for immunohistochemistry replication studies. We collected the 47 DLBCL cases, used in an earlier study, in the tissue bank of the department of Pathology UMCG between 1999 and 2012 based on the same criteria mentioned above. 14 The study protocol was consistent with international ethical and
28 29
on Harmonization Guidelines for Good Clinical Practice). The same patient cohort was used in an earlier study 14, and according to the Local ethics review board of
the Pathology department of the University Medical Centre Groningen fulfilled requirements for patient anonymity and were in accordance with their regulations. The Medical ethics review board waives the need for approval if rest material is used under law in the Netherlands, and waives the need for informed consent when patient anonymity is assured.
Immunohistochemistry
Formalin fixed, paraffin embedded (FFPE) tissues were stained for CD10 (Rabbit, clone SP67), BCL6 (mouse, clone CI191E/A8), MUM1 (mouse, clone MUM-1p) by Ventana (Roche, Tucson, USA) and were considered positive when >30% of the tumour cells stained positive. FoxP1 (Abcam, Cambridge, UK), MYC (Ventana, Roche, Tucson, USA) was considered positive when >60% and >80% of the tumour cells stained positive. For validation of the proteomics results formalin fixed, paraffin embedded (FFPE) tissue sections were stained for GLMN (ab170776 (Abcam, Cambridge, UK)), ADK (HPA038391, Sigma-Aldrich, Darmstadt, Germany), ARMC6 (HPA041420, Sigma-Aldrich, Darmstadt, Germany) and RPL23 (HPA003373, Table 1 Features of DLBCL NOS cases used for super-SILAC
Cases Hans Age Sex Ann Arbor stage IPI 1 non-GCB 51 male 3 1 2 non-GCB 78 male 3 4 3 non-GCB 77 male 1 2 4 non-GCB 61 female 3 2 5 non-GCB 66 male 4 3 6 non-GCB 71 male 1 1 7 GCB 66 male 3 2 8 GCB 14 male 1 unknown 9 GCB 48 female 4 3 10 GCB 53 female 4 2 11 GCB 58 male 1 unknown 12 non-GCB 78 male 4 3
28 Sigma-Aldrich). For western blotting the same antibodies were used. For all 29
stainings, reactive tonsils and other external controls were used as indicated by the manufacturer, publications on these antibodies, as well as the Protein Atlas (https:// www.proteinatlas.org/). Additionally, in all individual slides staining patterns of cells in the microenvironment were used as internal negative and positive controls.
The cases were classified based on the Hans algorithm using CD10, BCL6 and IRF4/ MUM1 (cut off level 30% of the tumour cells) into either non-GCB DLBCL or GCB DLBCL (table 1 and 2). 4 For validation and replication analyses of the proteomics
results also the Visco 3-thiered algorithm with CD10, FoxP1 and BCL6 (cut off levels 30%, 60% and 30%) was used.
Formalin-fixed paraffin-embedded (FFPE) tissue sections were stained for GLMN, ADK, ARMC6 and RPL23 (table S1). All immunohistochemical stainings were independently scored by LM and PK and discrepancies were discussed at a multi-head microscope. Tumour cells were scored in four categories using predefined thresholds based on the number of moderate to strong positive cells: 0%, 1-40%, 41-70% and 71-100%. Thus, cases with a very weak staining in all tumour cells were scored as 0%. In all cases a low percentage of the cells should show some staining, completely blank cases were considered as non-determinable. For statistical analysis Table 2 Clinical features of the replication series of DLBCL NOS
Characteristics non-GCB (20) GCB (27) male (%) 11 (55) 15 (56) median age (range) 66 (9-85) 56 (20-76) Ann Arbor Stage
I/II (%) 4 (20) 12 (44) III/IV (%) 14 (70) 9 (33) unknown (%) 2 (10) 6 (23) IPI 0-1 (%) 2 (10) 11 (41) 2-3 (%) 10 (50) 7 (26) 4-5 (%) 5 (25) 1 (4) unknown (%) 3 (15) 8 (29) * According to Hans classifi cation
30 31
we made two categories: cases with 0-40% positive tumour cells were considered as “negative” and cases with 41-100% positive tumour cells as “positive”. In rare borderline cases with a heterogeneous staining intensity, a very strong staining intensity of a subpopulation of tumour cells led to a positive categorization. Purification of tumour cells
Tumour cells were isolated using Dynabeads® CD19 Pan B (number 11143D,
Thermo Fisher Scientific, Waltham MA, USA) combined with DETACHaBEAD® CD19 (number 12506D, Thermo Fisher Scientific, Waltham MA, USA). Subsequently, the cell suspensions were depleted of naïve B-cells using anti-IgD coated Dynabeads (the tumour cells were IgD negative as assessed before by immunohistochemistry on frozen tissue sections). After this procedure the purity of all samples was checked by flow cytometry for expression of CD20, κ and (IQ Products, Groningen, The Netherlands). After purification, cells were washed three times with cold PBS and centrifuged at 1200 rpm for 5 minutes at 4 °C and subsequently lysed in lysis buffer (Cell signalling technologies, Danvers, USA, #9803) and placed on ice for 30-45 min. The supernatant containing membrane and cytoplasmic proteins was collected by centrifugation at 14.000 rpm for 10 minutes at 4 °C. The final protein concentration was measured using a BCA protein assay (Bio-Rad, Hercules CA, USA).
Culturing cell lines for super-SILAC
Eight B-cell lymphoma cell lines (DSMZ, Braunschweig, Germany) were selected for generating the super-SILAC reference sample; SC-1 , DoHH2 (transformed follicular lymphoma), OCI-LY3, U2932 (non-GCB DLBCL), DHL-5, DHL-6, SU-DHL-10, and SU-DHL-4 (GCB DLBCL). The cell lines were cultured in RPMI (Thermo Fisher) and penicillin/streptomycin (P/S) with either 10% fetal bovine serum (FBS) (OCILY3, SC-1, DoHH2 and SU-DHL-4) or 20% FBS (SU-DHL-5, SU-DHL-6, SU-DHL-10) supplemented with heavy 13C
6 L-Lysine-2HCl (Thermo Scientific, prod #88431) and 13C
615N4 L-Arginine-HCl (Thermo Scientific prod #88434). All cell lines were tested
negative for mycoplasma. Cells were cultured for approximately 10 cell passages to allow maximum incorporation of the labelled amino acids in all proteins. 15,16 The
incorporation was checked with mass spectrometry for each individual cell line. After confirmation of sufficient incorporation, all cell lines were lysed in lysis buffer (Cell signalling technologies, #9803) and 20-fold concentrated with the Vivaspin® 2 Centrifugal Concentrator. The final protein concentration was measured using a BCA protein assay (Bio-Rad, Hercules CA, USA).
30 31
Super-SILAC
The cell lines were mixed at equal protein amounts and used as an internal standard, to allow relative quantification of the protein amount and allow direct comparison of the protein levels obtained for each primary DLBCL sample. Each unlabelled patient sample was mixed 1:1 with the super-SILAC mix (50µg:50µg protein). Samples were heated for 5 min at 100°C before SDS-PAGE gel electrophoresis (2 hours, 110V). Each protein lane was divided into 15 equal pieces. No extra reverse phase purification was done. Each piece was washed twice in 400 µl MilliQ water for 15 min and washed twice in ± 400 µl 50% acetonitrile (ACN). Then 400 µl 100% ACN was added and incubated for 15 min. After removing ACN, ± 100 µl 10 mM dithiothreitol (DTT) (made in 100 mM ammonium bicarbonate pH 8-8.5) was added and incubated for 1 hour at room temperature. DTT was discarded and the gel pieces were covered with 100 µl 55 mM iodoacetamide (made in 100 mM ammonium bicarbonate pH 8-8.5) and incubated for 45 min at room temperature in the dark. The gel pieces were washed once with 400 µl MilliQ for 15 min and afterwards washed twice with 400 µl 50% ACN for 15 min. Then the gel pieces were washed once with ± 400 µl 100% ACN for 15 min. The 100% ACN was discarded and 40 µl 5 ng/µl trypsin solution (made in 20 mM ammonium bicarbonate pH 8-8.5) was added. The gel pieces were re-hydrated in trypsin solution for 15 min. The trypsin solution was removed and 50 µl 20 mM ammonium bicarbonate pH 8-8.5 was added to cover the gel pieces fully and digested overnight at 37°C. To extract the peptides 1 µl 100% formic acid added for 5 min (on shaker). Finally the gel pieces were centrifuged for 1 min on 5000rpm and the liquid was collected and gel pieces were discarded.
Protein identification
All samples were analysed on the Orbitrap LC-MS (Thermo Fisher Scientific, Waltham MA, USA) and with the PEAKS proteomics software platform. ProID 1.1 software (Applied Biosystems, Foster City, CA) 17 was used to identify proteins from the MS
datasets according to Swiss-Prot database. 18 Only proteins with a -10IgP of at least 50
and coverage by at least 2 peptides were considered. The list with predicted proteins was collapsed to generate a list with unique proteins. The differences in protein expression levels are indicated by the heavy/light ratio of each protein relative to the protein amount in the super-SILAC mix.
Data analysis of non-GCB DLBCL and GCB DLBCL
Data analysis was done with log2-transformed ratios, without normalization or baseline transformation, using GeneSpring GX software (version 14.9, Agilent
32 33
Genomics, Santa Clara CA, USA). Missing protein ratios were left blank. Proteins were filtered with the criterion that at least 6 out of 7 conditions for the non-GCB DLBCL group or 4 out of 5 conditions for the GCB DLBCL group should have expression values above the background.
Western Blot
Twenty million cells were washed with PBS and lysed in RIPA buffer (50 mM Tris/ 150 mM NaCl/ 2.5 mM Na2EDTA/1% Triton X-100, 0.5% mM sodium deoxycholate/0.1% SDS in dH20) with 1 mM phenylmethanesulphonyl fluoride for 30-45 minutes on ice. Protein concentration was determined using the Pierce™ BCA Protein Assay Kit (#23227; Thermo Scientific, Waltham MA, USA). Samples were loaded at 40µg per lane and electrophoresis and blotting was performed according to standard protocols. The antibodies used for immunohistochemistry were also used for western blotting as described in the supplementary material. Staining with primary antibodies for ARMC6, GLMN and RPL23 was done overnight and staining for GAPDH (1:20,000; clone 6C5 cat nr. 600-502, Novus bio, Centennial CO, USA) was done for one 1 hour at 4°C. Statistical analysis
A student’s t-test without multiple testing correction was performed to identify differentially expressed proteins in the proteomics data. In addition, we applied a 4-fold difference to select the most promising candidates. The chi-square test for trend was used (i.e. linear-by-linear association test) for validation and replication immunohistochemistry. P-values <0.05 were considered significant.
RESULTS
Selection of cases used for proteomics
Viably frozen cells of the 12 cases (5 GCB and 7 non-GCB DLBCL cases) were
successfully purified, and the purity of the tumour cell fraction as determined by CD20 varied between 85 and 99% (table S2). The purity as assessed for surface kappa/ lambda immunoglobulin expression in flow cytometry was more difficult to interpret due to high unspecific background staining because of we could not perform stringent washing procedures for fragile tumour cells, and in two cases due to the presence of many apoptotic cells. Based on the combined analysis, we estimated the percentage of polyclonal B-cells in the purified tumour cells used for proteomics to range from 0 to 45%, with a median of 8% (table S2).
32 33
Proteomics analysis
The total number of unique proteins identified in the patient samples was 5627. The number of proteins detected per sample ranged from 2273 to 3154. After filtering for proteins detected in at least 4/5 GCB or 6/7 non-GCB DLBCL cases 2059 proteins remained for further analysis. A significant difference between GCB and non-GCB DLBCL was observed for 132 proteins, of which 37 proteins showed at least a 4-fold difference, 24 proteins at least 5 fold and 10 proteins at least 6 fold (figure 1). Of the three proteins used in the Hans algorithm, only MUM1/IRF4 was included in the list of proteins with a slightly higher ratio in non-GCB DLBCL patient samples. BCL6, which is almost exclusively localized in the nucleus, is missing since we did not analyse nuclear proteins, and CD10 was not found. Four proteins (RPL23 (FC 48,4), ARMC6 (FC 10,4), STX4 (FC 6,9), XPNP1 (FC 6,2), showed higher levels in non-GCB DLBCL cases. The other six proteins, respectively GLB1 (FC 32,9), GLMN (FC 15,2), ADK (FC 10,5), PSMG4 (FC 8,2), PSAP (FC 6,5), PHPT1 (FC 6,2), , showed higher levels in GCB DLBCL cases (Fig 1). We selected 4 proteins for immunohistochemical validation and replication based on their fold change and the availability of antibodies suitable for immunohistochemistry. Ribosomal Protein L23 (RPL23) showed the most pronounced difference with more than 48-fold higher ratios in non-GCB DLBCL (range 834 to 37343), while GCB DLBCL cases showed low to very low expression levels (70 to 318). Armadillo repeat-containing protein 6 (ARMC6) levels showed 10-fold higher ratios in non-GCB DLBCL (273 to 14948). In GCB DLBCL cases the ratio varied between 168 to 1396. Glomulin (GLMN) showed >15 fold higher in GCB DLBCL cases (4739 to 12188) compared to non-GCB (0-14596). ADK ratios were 10-fold higher in GCB DLBCL, albeit with marked heterogeneity: low levels in 2 patient samples (856 and 2360) and high ratios in the 3 other patients (9063, 7886 and 7334). The ratios in non-GCB DLBCL patients varied between 93 and 1184 (table S3). GLB1, shows a high (32,9) fold change between both groups. We explored this protein immunohistochemically, unfortunately the staining showed a very dominant nuclear pattern (while it should mainly stain cytoplasmic compartments). Therefore we were unable to validate its expression by immunohistochemistry.
Western Blot
Western blot analysis was performed to check the correct molecular weight as detected by the selected antibodies and to check that the proteins were expressed in the 8 cell lines used for the super-SILAC mix. The purpose of the western blot was primarily to check that the proteins we tested for, were expressed in the cell line panel we used for normalisation, that with the same antibodies as used for
34 35
immunohistochemistry. In particular we wanted to check which isotype, either 48 kDa or 68kDa of GLMN was present. As the antibody cannot distinguish between FAP48 and FAP68, which have different functions. GLMN was expressed in all cell lines except for OCI-LY3 and showed a molecular weight of 68kDa. ARMC6 was expressed in all cell lines. RPL23 protein was detected in all cell lines but DoHH2 (figure 2). Validation on 10 of the super-SILAC cases
Validation of the proteomics results by immunohistochemistry was performed on 10 of the 12 super-SILAC cases (figures 3 and 4). The paraffin blocks of two other cases did not contain sufficient tissue for additional stainings. Overall, we observed trends that were similar to the proteomics results for all four proteins, but due to the low numbers of cases no definitive conclusions could be drawn.
Figure 1 Overview of the super-SILAC ratios for the 4 proteins (from high to low fold change), RPL23 (FC 48,4), ARMC6 (FC 10,4), STX4 (FC 6,9), XPNP1 (FC 6,2)) with higher expression in non-GCB and for the 6 proteins (from low to high fold change)(PHPT1 (FC 6,2), PSAP (FC 6,5), PSMG4 (FC 8,2), ADK (FC 10,5), GLMN (FC 15,2), GLB1 (FC 32,9)), with higher expression in GCB DLBCL with on the Y-axis SILAC ratio (FC, fold change).
34 35
Immunohistochemical staining of the replication cohort
Replication was done on 47 independent DLBCL cases (figures 3 and 4). Twenty cases were subtyped as non-GCB DLBCL and 27 cases as GCB DLBCL. RPL23
showed significantly more positive cases (p = 0.0234) in non-GCB (89%) as compared to the GCB DLBCL cases (58%). ARMC6 expression was observed at a higher
frequency in non-GCB than in the GCB DLBCL group, albeit not significantly. Immunohistochemical validation of GLMN showed a significant difference (p = 0.005) with considerably more positive cases in GCB DLBCL (92%) than in non-GCB DLBCL (55%). ADK expression was not different in our replication series. Two of 4 proteins showed a significant difference consistent with the proteomics findings and discriminated between GCB and non-GCB DLBCL as defined by the Hans algorithm. In addition to the Hans algorithm we also used the Visco 3-thiered algorithm for the immunohistochemical validation and replication analysis. Because of some drop outs, we combined the validation and replication series. Very similar results were seen using the Hans and Visco algorithm (see supplementary figure 3).
Figure 2 Western blot of three differentially expressed proteins in the cell lines used for to generate the super-SILAC protein mix. ARMC6, RPL23, GLMN, for 8 DLBCL cell lines, from left to right: DoHH2, SU-DHL-4, SC-1, SU-DHL-5, SU-DHL-6, SU-DHL-10, U2932, OCI-LY3.
36 37
Comparison of peptide fragments identified by proteomic analysis and immunohistochemical antibodies
We compared the peptide fragments identified by proteomic analysis to the epitope of the immunohistochemical antibodies used. We checked the peptides and for all proteins peptides along the whole protein were found. We used mainly polyclonal antibodies: the antibody against GLMN was raised against aa505-533: 2 peptides were found in our cases. The antibody against RPL23 was raised to aa10-88: several peptides were found in this area in our cases. The antibody against ARMC6 was raised against aa 44-140: several peptides were found in this area. The antibody for ADK was raised to aa89-170: several peptides were found in our cases.
Comparison with individual markers of the Hans criteria
To determine whether any of the markers identified in this study correlates with the individual markers used in the Hans algorithm, we investigated the total cohort of 58 cases (validation plus replication) (figure S1). However not significant, there was a Figure 3 Representative images of the immunohistochemical staining results of the four selected proteins in GCB and non-GCB DLBCL. A: ARMC6, B: RPL23, C: GLMN, D: ADK. ARMC6 (non-GCB case), GLMN (GCB case) and RPL23 (non-GCB case) showed cytoplasmic expression and ADK (GCB case) staining was observed in the nucleus.
36 37
Figure 4 Results of the immunohistochemistry of the 4 selected proteins (RPL23, ARMC6, GLMN and ADK). A-D: results of 10 cases with sufficient material for
validation and E-H: results of the replication cohort of 47 independent DLBCL cases. trend seen for RPL23 (p = 0.0578) and ARMC6 (p = 0.0730) expression in correlation with negative CD10 cases. Negative immunohistochemical CD10 expression is a prerequisite for non-GCB DLBCL. RPL23 and ARMC6 expression were both upregulated in non-GCB DLBCL in our proteomic analysis.
38 39
DISCUSSION
Using super-SILAC on membrane and cytoplasmic proteins of purified tumour cells isolated from primary viable cell suspensions we uncovered several proteins that were significantly differentially expressed between GCB and non-GCB DLBCL. We found no expression of CD10, similar to Deeb et al. who found only one peptide of CD10. 19 The lack of BCL6 in our analysis was likely due to the fact that the protein is
almost exclusively localized in the nucleus, a compartment not investigated by us. Our analysis showed some differential expression of MUM1/IRF4, with slightly higher levels in non-GCB DLBCL. This may be explained by the fact that MUM1/IRF4 is localized within the nucleus and cytoplasm.
Two out of four proteins, i.e. RPL23 and GLMN, selected for validation and replication by immunohistochemistry showed a significantly differential expression pattern consistent with the super-SILAC results. The total number of unique proteins identified in our study were similar to those in other proteomics studies. 9-10,20
Deeb et al. generated a super-SILAC mix with nine B-cell lymphoma lines and identified 6263 proteins in a super-SILAC analysis on 5 non-GCB DLBCL cell lines and 5 GCB DLBCL cell lines. They established a signature of 55 proteins that could differentiate between non-GCB DLBCL and GCB DLBCL subtypes. 9 Twenty-five of
the 55 proteins overlapped with the proteins found in our study. In a second study Deeb et al. applied super-SILAC to differentiate between non-GCB DLBCL and GCB DLBCL using 20 FFPE DLBCL tissue samples. This revealed 5480 proteins with 343 differentially expressed proteins. The overlap with our list of 1975 differentially expressed proteins was 74 out of 343 proteins. 10 Ruetschi et al. analysed frozen tissue
sections of five relapsed and five long-term progression-free DLBCL patients using super-SILAC. The reference super-SILAC mix consisted of four DLBCL cell lines and one Burkitt lymphoma cell line. Of the 3588 proteins, 87 proteins were differentially expressed. Fifty-five of the 87 proteins were found in our differentially expressed list.
11 Two proteins within these 3 studies were similar: CD44 and MUM1/IRF4.
We followed a two-step approach to validate our findings with
immunohistochemistry, validation in the same cases and replication in an
independent group consisting of 20 non-GCB and 27 cases GCB DLBCL. We used The Hans algorithm and Visco algorithm to classify our cases. The Hans algorithm is the most commonly used approach to classify DLBCL cases in the routine diagnostic setting. 8 According to Visco et al. the Hans, Choi and Tally algorithms performed
almost equally well and the 3- and 4-thiered Visco algorithms somewhat better. 21 We
38 39
the immunohistochemical validation and replication analyses of the four remaining proteins. For 7 cases FoxP1 immunohistochemistry was not present because there was no tissue available. Three cases switched for non-GCB to GCB and vice versa. Only minor differences were found between both algorithms mainly due to a smaller series used for the Visco algorithm as can be seen in supplemental figure 3.
The data on the validation cohort showed staining patterns that were consistent with the proteomics data, but the cohort was too small to perform meaningful statistical testing.
In the replication series significantly different expression that correlated with proteomics were found for 2 of 4 selected proteins, i.e. RPL23 and GLMN. RPL23 showed a more frequent staining in non-GCB than GCB DLBCL by both approaches. Meng et al. studied the RAS–RPL23–MDM2–p53 pathway, and showed that increased levels of RPL23 induced by RAS were associated with increased p53 expression levels. 22 We investigated the relationship with p53 expression in the
tumour cells of 36 cases that had been stained in our series and observed a similar, albeit not significant, trend (figure S2a). RPL23 can be induced upon activation of MYC, leading to a positive feedback loop. MYC overexpression is observed in a considerable part of DLBCL, in particular in non-GCB DLBCL 23, and it is associated
with a poor survival, in particular if combined with BCL2 protein overexpression. 23-24 Qi et al. observed MYC overexpression in combination with high levels of RPL23
in SKM-1, an acute myeloid leukaemia cell line. 25 We therefore performed MYC
immunohistochemistry and found a higher immunohistochemical expression of MYC in RPL23 positive cases (figure S2b). In the literature 23 MYC overexpression was
especially observed in non-GCB cases. As shown in figure S2c we could not support this result. In view of the positive association between MYC and RPL23 expression in our cases it might be suggested that RPL23 is regulated by MYC in DLBCL (figure S2). However, we did not perform any functional studies to support this. Expression of GLMN was much more frequently observed in GCB DLBCL cases, both in the proteomics analysis and the replication study. GLMN is a FK506-binding protein (FKBP) associated protein, with two potential isoforms also known as FKBP associated protein 48 (FAP48) or FKBP associated protein 68 (FAP68). 26-27 In the DLBCL cell
lines, we only observed the longer 68 kDa isoform. GLMN is part of the Skp1-Cullin-F-box-like complex and plays a role in the differentiation of smooth muscle cells. GLMN loss of function mutations in vascular smooth-muscle cells results in increased angiogenesis. 26,28 We observed a relatively infrequent expression in non-GCB DLBCL,
40 41
angiogenesis as assessed by micro vessel density measurements in non-GCB DLBCL compared to GCB-DLBCL, which was linked to poor clinical outcome in non-GCB DLBCL. 29-30 In conclusion, we performed super-SILAC on purified primary DLBCL
tumour cells and showed a consistent differential expression pattern of two proteins between GCB and non-GCB type DLBCL.
ACKNOWLEDGEMENTS
- We thank M.J. van Schaijk for critical reading of the manuscript and helpful suggestions for improving the comprehensibility of the paper.
- We thank M.P. de Vries and H.P. Permentier (Interfaculty Mass Spectrometry Centre, University of Groningen) for their technical assistance and helpful advice.
- This project was supported by the Dutch Cancer Society under grant number KWF RUG 2011-5252.
CONFLICT OF INTEREST
40 41
REFERENCES
1. Lenz G, Wright GW, Emre NCT, Kohlhammer H, Dave SS, Davis RE, et al. Molecular subtypes of diffuse large B-cell lymphoma arise by distinct genetic pathways. Proc Natl Acad Sci USA. 2008 Sep 9;105(36):13520–5.
2. Lenz G, Wright G, Dave SS, Xiao W, Powell J, Zhao H, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med. 2008 Nov 27;359(22):2313–23.
3. Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature. 2000 Feb 3;403(6769):503–11.
4. Hans CP. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004 Jan 1;103(1):275–82.
5. Choi WWL, Weisenburger DD, Greiner TC, Piris MA, Banham AH, Delabie J, et al. A New Immunostain Algorithm Classifies Diffuse Large B-Cell Lymphoma into Molecular Subtypes with High Accuracy. Clinical Cancer Research. 2009 Aug 31;15(17):5494–502.
6. Meyer PN, Fu K, Greiner TC, Smith LM, Delabie J, Gascoyne RD, et al. Immunohistochemical Methods for Predicting Cell of Origin and Survival in Patients With Diffuse Large B-Cell Lymphoma Treated With Rituximab. Journal of Clinical Oncology. 2011 Jan 6;29(2):200–7.
7. Gutierrez-Garcia G, Cardesa-Salzmann T, Climent F, Gonzalez-Barca E, Mercadal S, Mate JL, et al. Gene-expression profiling and not immunophenotypic algorithms predicts prognosis in patients with diffuse large B-cell lymphoma treated with immunochemotherapy. Blood. 2011 May 5;117(18):4836–43.
8. Coupland SE. The challenge of the microenvironment in B-cell lymphomas. Histopathology. 2011 Jan 24;58(1):69–80.
9. Deeb SJ, D’Souza RCJ, Cox J, Schmidt-Supprian M, Mann M. Super-SILAC allows classification of diffuse large B-cell lymphoma subtypes by their protein expression profiles. Molecular & Cellular Proteomics. 2012 Feb 22;11(5):77–89.
10. Deeb SJ, Tyanova S, Hummel M, Schmidt-Supprian M, Cox J, Mann M. Machine Learning-based Classification of Diffuse Large B-cell Lymphoma Patients by Their Protein Expression Profiles. Molecular & Cellular Proteomics. 2015 Nov 1;14(11):2947–60.
11. Rüetschi U, Stenson M, Hasselblom S, Nilsson-Ehle H, Hansson U, Fagman H, et al. SILAC-Based Quantitative Proteomic Analysis of Diffuse Large B-Cell Lymphoma Patients. International Journal of Proteomics. 2015;2015:1–12.
42 43
12. Geiger T, Cox J, Ostasiewicz P, Wisniewski JR, Mann M. Super-SILAC mix for quantitative proteomics of human tumor tissue. Nature Methods. 2010 Apr 4;7(5):383–5.
13. Swerdlow SH, Campo EB, Harris NL, Jaffe ES, Pileri SA, Stein H, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. 4 ed. Vol. 2. 2017.
14. van der Meeren LE, Visser L, Diepstra A, Nijland M, van den Berg A, Kluin PM. Combined loss of HLA I and HLA II expression is more common in the non-GCB type of diffuse large B cell lymphoma. Histopathology. 2018;72(5):886–8.
15. Ong S-E, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 2002 May;1(5):376–86. 16. Ong S-E, Mann M. A practical recipe for stable isotope labeling by amino acids in
cell culture (SILAC). Nat Protoc. 2007 Jan;1(6):2650–60.
17. Tang WH, Halpern BR, Shilov IV, Seymour SL, Keating SP, Loboda A, et al. Discovering Known and Unanticipated Protein Modifications Using MS/MS Database Searching. Anal Chem. 2005 Jul;77(13):3931–46.
18. Boeckmann B. The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003. Nucleic Acids Research. 2003 Jan 1;31(1):365–70.
19. Deeb SJ, D’Souza RCJ, Cox J, Schmidt-Supprian M, Mann M. Super-SILAC Allows Classification of Diffuse Large B-cell Lymphoma Subtypes by Their Protein Expression Profiles. Mol Cell Proteomics. 2012 May 14;11(5):77–89.
20. Huang X, Shen Y, Liu M, Bi C, Jiang C, Iqbal J, et al. Quantitative Proteomics Reveals that miR-155 Regulates the PI3K-AKT Pathway in Diffuse Large B-Cell Lymphoma. The American Journal of Pathology. 2012 Jul;181(1):26–33.
21. Visco C, Li Y, Xu-Monette ZY, Miranda RN, Green TM, Tzankov A, et al. Comprehensive gene expression profiling and immunohistochemical studies support application of immunophenotypic algorithm for molecular subtype classification in diffuse large B-cell lymphoma: a report from the International DLBCL Rituximab-CHOP Consortium Program Study. Nature Publishing Group; 2012 Aug 8;26(9):2103–13.
22. Meng X, Tackmann NR, Liu S, Yang J, Dong J, Wu C, et al. RPL23 Links Oncogenic RAS Signaling to p53-Mediated Tumor Suppression. Cancer Research. 2016 Aug 31;76(17):5030–9.
