HDAC Inhibitors Sensitize Colon Cancer Stem Cells (Cscs) to Apoptosis Via FOXO4 Regulation

UvA-DARE (Digital Academic Repository)

Crossing Borders: Manipulating stemness through carcinogenesis

Zimberlin, C.D.

Publication date 2016 Document Version Final published version

Link to publication

Citation for published version (APA): Zimberlin, C. D. (2016). Crossing Borders: Manipulating stemness through carcinogenesis.

General rights 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), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

Download date:02 Oct 2021 8 HDAC inhibitors sensitize colon cancer stem cells (CSCs) to apoptosis via FOXO4 regulation

Manuscript in preparation

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 143143 008/05/168/05/16 22:0522:05 Chapter 8

HDAC inhibitors sensitize colon cancer stem cells (CSCs) to apoptosis via FOXO4 regulation

C.D.Zimberlin*1, S.Colak*1,2, M.S.Roca*1,3, K.Cameron1, E.E.Santo4, S.R.van Hooff1, H.Rodermond1, S.Simmini1, M.Bots1, C.M.Grandela1,5, B.M.Burgering6,7, J.P.Medema1,7

* These authors contributed equally

1 LEXOR (Laboratory of Experimental Oncology and Radiobiology), Center for Experimental Molecular Medicine, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands

2 Department of Molecular Cell Biology, Leiden University Medical Center, Einthovenweg 20, 2333 ZC, Leiden, The Netherlands

3 Experimental Pharmacology Unit, Istituto Nazionale Tumori Fondazione G. Pascale - IRCCS, 80131 Naples, Italy

8 4 Department of Pathology and Laboratory Medicine , Weill Cornell Medicine , New York City , NY , USA

5 Current address: Pluriomics BV, Leiden 2333BD, The Netherlands

6 Department of Molecular Cancer Research, University Medical Center Utrecht, Universiteitsweg 100, 3584 CG, Utrecht, the Netherlands

7 Cancer Genomics Center, The Netherlands

Key words: Histone deacetylase, Tumor initiating cells (TIC), LBH-589, BCL-XL, FOXO, ABT-737, Colorectal cancer, HDACi

Running title: HDAC inhibitors target CSCs for cell death

144

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 144144 008/05/168/05/16 22:0522:05 Chapter 8

Abstract

In many cancer types cancer stem cells (CSCs) have been identifi ed and suggested to fuel tumor growth. CSCs are reportedly more resistant to chemotherapy and radiotherapy, providing a mechanism for disease relapse after therapy. Targeting CSC therefore provides a possible therapeutic window for the treatment of tumors. Using a single cell-based analysis of cell death and a small-scale compound screen, we studied sensitization of colon-CSCs towards conventional chemotherapy. We identifi ed HDAC inhibitors (HDACi) as potential candidates, which strongly enhanced oxaliplatin-induced apoptosis in colon-CSCs. Closer examination revealed that HDACi induced a change in cellular morphology and a striking increase in the level of differentiation markers with a concomitant decrease of stem cell markers. Importantly, de novo transcription was required for sensitization and subsequent gene-expression analysis pointed to the involvement of FOXO transcription factors, particularly FOXO4. Treatment with HDACi consistently increased FOXO4 expression levels and lowered the expression of BCL-XL. Importantly, knockdown of FOXO4 prevented this HDACi-induced decrease in BCL-XL, indicating that FOXO4 induction is instrumental in this regulation. More importantly, we fi nd that HDACi specifi cally decreased the mitochondrial apoptotic threshold in colon-CSCs as measured with ABT-737. Our results therefore show that HDACi sensitize colon-CSCs through a novel mechanism involving the loss of stemness and a FOXO4-dependent lowering of the apoptotic threshold. Both activities provide a rational for the combination of HDACi with chemotherapy in colorectal cancer. 8

145

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 145145 008/05/168/05/16 22:0522:05 Chapter 8

Introduction

Colorectal cancer (CRC) is one of the leading causes of cancer-related mortality worldwide. Therapeutic developments have increased 5-year patient survival rates to ±90%, however, once distance metastasis has occurred survival drops to less than 15% [1]. Patients with advanced stage CRC are routinely treated with a combination of ﬂ uoropyrimidines (Capecitabine or Fluorouracil) and oxaliplatin or irinotecan. Despite the development of targeted therapies, expanding experimental evidence indicates robust signaling redundancy present within CRC lesions, leading to acquired secondary resistance [2]. Research efforts to efﬁ ciently block tumor growth and adaptation should thus be prioritized in the development of new therapeutic strategies.

CRC lesions are reported to contain a cellular hierarchy, of which the colon cancer stem cells (CSCs) are at the apex. The CSCs are a small subset of cells, which drive tumor growth and progression. Colon-CSCs can be deﬁ ned by a variety of different markers including Wnt pathway activity, LGR5, OLFM4 and CD133 (reviewed in [3]). Intriguingly, despite being proliferative, colon-CSCs been have been proven to display selective resistance to conventional chemotherapy, thus escaping chemotherapeutic insults and allowing for a repopulation of the tumor [4-6].

The resistance of CSCs appears to be an intrinsic property that is lost upon differentiation and exploiting the differences between CSCs and their more differentiated progeny may thus be the key to ﬁ nding new ways to sensitize CSCs to chemotherapeutics. Indeed upon activation of the 8 Notch pathway using a DLL4 neutralizing antibody in vivo, CSCs undergo differentiation and can subsequently be targeted by chemotherapeutics [7]. Similarly, activation of the ER stress pathway [8] or exposure to BMP4 ligands [9] can induce differentiation of colon-CSCs and sensitize them to chemotherapy. Recently, we reported that, when compared to their differentiated progeny, colon-CSCs are less primed to mitochondrial cell death and can be sensitized by inhibiting the anti-apoptotic molecule BCL-XL, with small compound BH3 mimetics [4]. Further identiﬁ cation of compounds that share the ability to modulate CSC resistance could potentially provide key mechanistic insight into CSC functioning, which can subsequently be exploited for therapeutic use.

We therefore performed a small-scale screen of selected compounds and identiﬁ ed histone deacetylase inhibitors (HDACi) to effectively sensitize CSCs to chemotherapy, which involved de novo transcription/translation. Motif analysis of microarray data indicated a role for the FOXO family members, particulary FOXO4. HDACi upregulate FOXO4 expression, which negatively regulates BCL-XL levels in primary colon cells enriched for CSCs. In addition, we observed loss of stemness and increased differentiation-associated gene expression. Decreased levels of BCL- XL lowered the apoptotic threshold of colon-CSCs sensitizing them to chemotherapeutics. The

146

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 146146 008/05/168/05/16 22:0522:05 Chapter 8

transcriptional regulation of FOXO4 to alter the apoptotic threshold of colon-CSCs reveals a novel means to target these cells and enhance their chemotherapy sensitivity.

Results

HDAC inhibitors can sensitize colon-CSCs to oxaliplatin treatment

CRC develops in a step-wise manner, where aberration of the Wnt pathway by mutations in APC or CTNNB1 is viewed as an early event [10, 11]. Despite deregulation of the Wnt pathway in all CRC cells a heterogeneous expression of Wnt activity can be observed within a CRC lesion [12-14]. Previously we have shown that the heterogeneity in Wnt pathway activity can help defi ne different subset of cells, where high Wnt activity defi nes the clonogenic colon-CSC population and low Wnt activity defi nes their more differentiated progeny [12]. Using a single cell cloned primary spheroid culture, Co01, containing a TOP-GFP Wnt reporter, the Wnt pathway activity can be monitored by GFP expression. In this manner both differentiated cells (10% GFPlow) and CSCs (10% GFPhigh) can be examined within the same population. Recently we have described a fl ow cytometry technique, in which apoptosis based on activated caspase-3 can be measured in single cells allowing for an analysis of chemotherapy-induced cell death of differentiated and colon-CSCs under the same conditions. Based on this technique we showed colon-CSCs to be less sensitive to conventional chemotherapeutics than their differentiated counterparts due to a BCL-XL dependent roadblock in apoptotic signaling [4]. As BCL-XL targeting mimetics display relatively high toxicity we here opted for a strategy to fi nd novel means to sensitize colon-CSCs to oxaliplatin. To this end we performed a small-scale drug screen using our singe 8 cell-based apoptosis assay, testing 50 different therapeutic agents targeting multiple cellular processes (fi gure 1a-c, supplementary table 1). Interestingly, the highest sensitization of CSCs to chemotherapeutics was observed with two independent HDAC inhibitors (fi gure 1b-c and supplementary table 1). The establishment of chemosensitivity in colon-CSCs upon HDACi treatment was validated and expanded to various pan-HDAC inhibitors including panobinostat (LBH-589), vorinostat (SAHA), MS-275, valproic acid (VPA), Trichostatin A (TSA) and sodium butyrate (NaB) (fi gure 1d). Colon-CSCs hereby appear to be dependent on HDAC activity to maintain their chemoresistant state. In order to validate the effect of HDAC inhibition on chemotherapy-effi cacy, in vivo xenograft experiments were performed using oxaliplatin and the HDACi panobinostat. Colon-CSCs were injected in subcutaneously in mice and allowed to grow to a size of 100mm3. As reported before oxaliplatin alone at these low concentrations provides no survival benefi t to the mice due to a selective resistance of CSCs [5, 15]. Similar results were observed for HDACi alone. However, the combination provided a strong survival benefi t and even induced a complete regression in 2 of the 8 mice (Figure 1e). HDAC inhibition in combination with oxaliplatin thus signifi cantly improves chemotherapy response in vitro and in vivo.

147

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 147147 008/05/168/05/16 22:0522:05 Chapter 8

A. 1. Treat Primary CRC spheriods 2. Cell isolation + staining 3. Analysis of cell death

Wnt low % DeDeadad CellsCells Diffrentiated cells

10% Differentiated GFPlow Cells

% Dead Cells

10% GFPhigh Wnt high Stem Cells Stem cells

B. C. Cell death in CSCs in single vs combined treatment with oxaliplatin 100 Dif Ranking Inhibitor Fold induction Class Inhibits CSC 1 Vorinostat 3.9 Epigenetics HDACs 80 2 Panobinostat 3.7 Epigenetics HDACs 3 17-DMAG 2.5 Transc/Transl HSP90 60 4 SB203580 2.1 Signalling p38 MAPK 5 Roscovitine 1.8 Signalling CDK 40 6 NU1025 1.8 DNA repair PARP 7 Pepstatin A 1.7 Proteases Aspartic proteinases 20

% Caspase-3 active cells 8 SCD1 inhibitor 1.6 Metabolism Fatty acid synthase 9 Olaparib 1.6 DNA repair PARP 0 CtrlOxali Oxali + Oxali + Oxali + Oxali + 10 2-aminopurine 1.5 Transc/Transl PKR Vorino Pano 17-DMAG SB D. E. 80 100 Ctrl Oxali Pano 60 80 Pano + Oxali

60 8 40 40

20 % surviving mice p = 0.02 20 % Caspase-3 active cells

0 0 Oxali -++ -+-+-+-+-+- 0 35 70 105 140 CtrlPano Vorino MS-275 TSA VPA NaB Days after first treatment

Figure 1: HDAC inhibitors can sensitize CSCs to oxaliplatin treatment A) overview of the screening system used. B) Flow cytometry analysis of activated caspase-3 shows the outcome of the induced chemosensitivity in colon-CSCs (red) and differentiated cells (blue) when oxaliplatin (Oxali) is combined with: Vorinostat (Vorino) or Panobinostat (Pano) or 17-DMAG or SB203580 (SB). C) A table representing the top ten compounds that established the greatest fold change in cell death in colon-CSCs speciﬁ cally when compared to oxaliplatin alone. A full list of all the dosages and fold change observed is provided in Sup. Table 1. D) Flow cytometry analysis of activated caspase-3 shows that multiple HDAC inhibitors sensitize colon-CSCs to cell death in combination with oxliplatin. Panobinostat (Pano), vorinostat (Vorino), MS-275, valproic acid (VPA), Trichostatin A (TSA) and sodium butyrate (NaB). E) Survival curves of colon- CSC xenografts treated with the vehicle (Ctrl), panbinostat (Pano), oxaliplatin (Oxali) and a combination of panobinostat and oxaliplatin (Pano + Oxali). The grey box shows the time-scale when mice were treated. n=6

148

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 148148 008/05/168/05/16 22:0522:05 Chapter 8

HDAC inhibitors induce chemosensitivity by transcriptional regulation

HDAC together with their counterpart’s histone acetylases (HATs) maintain the acetylation status of proteins within the cell. The acetylation status of proteins is crucial for the maintenance of homeostasis and a plethora of cellular processes. HDACs have been shown to induce transcriptional repression as deacetylation of lysine residues on histone proteins induces chromatin compaction [16, 17]. However, non-histone proteins, including transcription factors, are also regulated in their activity by HDAC-mediated deacetylation, which makes the effect of HDACi both pleiotropic and cell type dependent. Using Actinomycin D (ActD) and Cycloheximide (CHX) to block transcription and translation respectively, we did observe that HDACi-induced chemosensitization fully depended on de novo mRNA/protein synthesis (ﬁ gure 2a).

In order to gain further insight into the mechanism underlying HDACi sensitization of colon- CSCs, gene expression profi les were generated from control or panobinostat-treated colon-CSCs or differentiated cells. Using a previously described CSC signature [18], panobinostat-treated colon-CSCs were found to have a decreased resemblance to colon-CSCs, gaining similarity to the more-differentiated cell population (fi gure 2b). Importantly, upon closer inspection panobinostat treatment decreased the expression levels of selected colon CSC markers LGR5, CD133, ASCL2, and EPHB2, while it resulted in increased expression levels of the differentiation markers; cytokeratin 20 (KRT20) and CEACAM5 (fi gure 2c and supplementary table 2). This decrease in CSC markers and concomitant increase in differentiation markers suggests that HDAC inhibition induces a differentiation-like phenotype, which may sensitize CSCs to 8 conventional therapeutics. Consistent with these fi ndings, panobinostat treatment revealed marked morphological changes in Co01 cells (supplementary fi gure 1). This morphological differentiation was even more evident when spheroid cultures were grown in Matrigel, where panobinostat induced the formation of strongly polarized structures, comparable to spheroids treated with FCS, a known differentiation factor [5] (fi gure 2d). These distinctive polarization changes were present in 90% of the structures observed upon panobinostat treatment (fi gure 2d) and coincided with increased and polarized expression of the differentiation marker cytokeratin 20 (KRT20), both on protein (fi gure 2d) and mRNA expression (fi gure 2e).

Microarray data also showed a signifi cant decrease in stem cell markers, which was confi rmed using qPCR-based analysis of LGR5 (fi gure 2f). Moreover, panobinostat treatment also decreased surface expression of stem cell markers LGR5 and CD133 (fi gure 2g). Importantly, these effects are not unique to this primary colon cancer culture, but could further be extended to several primary CRC spheroid cultures (supplementary fi gure 2). Combined the decrease in stem cell-associated markers and increase in differentiation markers, supports the notion that HDACi induces colon-CSCs to undergo differentiation.

149

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 149149 008/05/168/05/16 22:0522:05 Chapter 8

A. B.

50 1. 2. CSC_Ctrl 40 3. 1. 30 2. Dif_Ctrl 3. 20 1. 2. CSC_Pano

% Caspase-3 active cells 10 -1 0 1 3. C. LGR5 CD133 EPHB2 0 ActD - + -+--+-- -+- 12.5 9.8 CHX - - +--+-+- - -+ 11.0 9.7 Ctrl Oxali Pano Oxali+Pano 12.0 9.6 10.5 9.5 11.5

D. Expression 9.4 Ctrl 100 10.0 11.0 9.3 80 Pano - + - + - + - + - + - + CSC Dif CSC Dif CSC Dif ASCL2 KRT20 CEACAM5 60 13.0

10.5 12.5 11 40 10.0 12.0 10 KRT20 11.5 9.5 20 11.0 9 FCS % non-polarized structures 9.0 10.5

0 10.0 8 Pano- + 8.5 9.5 - + - + - + - + - + - + CSC Dif CSC Dif CSC Dif KRT20 LGR5 E. 0.5 F. 15 G. H. ** Ig Ctrl Ctrl Pano * KRT20 ** 30 0.4

Pano 10 0.3 20

0.2

5 Counts 10 0.1 Clonogenic capacity (%)

8 Expression (relative to B2M) Expression (relative to B2M) KRT20 0 0 CD133 LGR5 0 Pano- + Pano- + Pano- + Figure 2: The chemosensitization effect of HDAC inhibitors depends on a change in the transcriptional program leading to increased differentiation and loss of stemness. A) Activated capase-3 activity present in the colon-CSC population after treatment with control (ctrl), oxaliplatin (Oxali) and panobinostat (pano) with and without the presence of cyclohexamide (CHX) or actinomycin D (ActD). B) Gene expression profi ling was performed on colon-CSCs and differentiated cells (Dif), treated with and without panobinostat (pano), cluster analysis was performed based on a previously described CSC signature [18]. C) Expression level changes in the microarray for several stem cell markers (LGR5, CD133, ASCL2, EPHB2) and differentiated markers (KRT20, CEACAM5). D) Immunohistochemical stainings show the changes in morphology between control (Ctrl) Co01 CRC spheroid cells and treatment with FCS or panobinostat (pano), increased levels of cytokeratin 20 (KRT20) expression can be observed and the amount of polarized structures is quantifi ed in the barchart to the right. Scale bar 50μM. Magnifi cation x40. qRT-PCR analysis shows E) increased differentiation marker KRT20 upon panobinostat treatment and F) decreased stemness marker LGR5 G) Flow cytometry shows decreased levels of associated stem cell markers CD133 and LGR5. H) The observed loss in stemness coincides with a loss in clonogenic capacity upon panobinostat treatment. n≥3. *p value ≤ 0.05 (t-test). **p value ≤ 0.01 (t-test).

150

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 150150 008/05/168/05/16 22:0522:05 Chapter 8

To functionally study whether HDACi indeed affect the innate stem cell properties of colon- CSCs we performed a limiting dilution assay and observed the clonogenic capacity of the Co01 (ﬁ gure 2h) and Co108 (data not shown) spheriod cultures to be signiﬁ cantly reduced upon panobinostat treatment. Combined these results indicate that HDACi treatment drives colon-CSCs towards a more differentiated-like phenotype, decreasing clonogenic potential and potentially alleviating the apoptotic roadblock that prevents chemotherapy-induced cell death.

HDAC inhibitors alter the transcriptional proﬁ le via the FOXO family

The ability of HDACi to reprogram cellular transcriptional activity is well-accepted. However, the exact mechanisms and specifi c transcriptional regulators defi ning these programs are cell type-dependent and pleiotropic in nature. To gain better insight into the chemosensitizing effect of HDACi, we aimed to identify which transcription factors are responsible for the HDACi-induced global changes in gene activation in colon-CSCs, using the so-called Search Tool for Occurrences of Regulatory Motifs (STORM) analysis [19]. The STORM analysis maps differential gene expression between treated and untreated populations and defi nes transcription factors responsible for these active promoter regions and revealed that the transcriptional targets of FOXO family member, FOXO4, were the most signifi cantly enriched upon panobinostat treatment (fi gure 3a). Intriguingly, analysis of the microarray gene expression profi les as well as subsequent qPCR analysis indicated a clear HDACi-induced increase in FOXO4 as well as FOXO1 expression, whereas FOXO3 did not appear to be regulated (fi gure 3b,c). FOXO family members have previously been reported to display feedback regulation upon themselves, where active FOXO3 enhanced FOXO1 expression [20]. Using an inducible overexpression 8 system for FOXO4a3, which due to the phosphorylation point mutations is constitutively active, we observed that FOXO1 was also induced by FOXO4, while FOXO3 was unaffected (supplementary fi gure 3), suggesting that part of the FOXO1 increase upon panobinostat treatment could be attributed to increased FOXO4 expression.

FOXO family members shuttle, in a phosphorylation regulated manner, from the cytoplasm to the nucleus, where they are transcriptionally active. In colon-CSCs we could only observe FOXO4 in the nucleus also upon induction by HDACi, suggesting that it is transcriptionally active (supplementary ﬁ gure 3). The current data thus point to a model in which HDACi treatment enriches for and activates transcriptional proﬁ les of FOXO family members, in particular FOXO4. In order to investigate whether the observed differentiation effects in primary CRC spheroid cultures can be attributed to FOXO4 regulation, we generated cultures either over-expressing or lacking FOXO4.

To this end primary Co01 cultures were transduced and single cell cloned with the previously mentioned doxycycline inducible FOXO4a3 over-expression (FOXO4a3-OE) construct.

151

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 151151 008/05/168/05/16 22:0522:05 Chapter 8

A. B. Matrix ID T.F. Name Enrichment (%) P-Value FOXO4 FOXO1 FOXO3 M00472 FOXO4 9,1 6,79E -07 7.0 11.0 6.5 M04267 FOXP3 9,8 5,29E -06

6.5 M01388 DLX5 6,9 1,41E -05 10.5 6.0

M04399 MEOX2 7,3 2,35E -05 6.0

10.0 M01324 OCTAMER 6,7 2,74E -05 5.5

Expression 5.5 M03090 POU3F3 6,6 3,28E -05 5.0 9.5

M00639 HNF6 6,6 5,03E -05 5.0 M00138 OCT1 6,5 0,000116144 Pano - + - + - + - + - + - + M04424 PDX1 5,9 0,000153445 CSC Dif CSC Dif CSC Dif M01340 BARX1 6,1 0,000156552

C. D. FOXO4 E.F. KRT20 H. 0.015 FOXO4 0.15 FOXO1 0.015 FOXO3 0.10 1.5 LGR5 1.0 40 ** ** * * * *** ** * 0.08 0.8 * 30 0.010 0.10 0.010 1.0 0.06 0.6 20 0.04 0.4 0.005 0.05 0.005 0.5 10 0.02 0.2 Expression (relative to B2M) Expression (relative to B2M) Expression (relative to B2M) Clonogenic capacity (%) Expression (relative to B2M) 0 Expression (relative to B2M) 0 0 Expression (relative to B2M) 0 0 0 0 Pano- + Pano- + Pano- + Pano -++ - Pano -++ - Pano-+ + - Ctrl-EV 4a3-OE Ctrl-EV 4a3-OE Ctrl-EV 4a3-OE FOXO4 LGR5 KRT20 G. I. 3 J. K. ** 4a3-OE_Ctrl 4a3-OE_Pano ** * * * * 1.0 10 2

0.5 1 5

KRT20 KRT20 Expression (relative to B2M) 0 Fold change in Expression 0 Fold change in Expression 0 Pano-+ + - Pano-+ + - Pano-+ + - DOX-+ - + DOX-+ - + DOX-+ - + L. shFOXO4_Ctrl shFOXO4_Pano shFOXO4_DOX shFOXO4_DOX+Pano 8 KRT20 KRT20 KRT20 KRT20 Figure 3: HDAC inhibitors regulate the transcriptional program to induce chemosensitization. A) Search Tool for Occurrences of Regulatory Motifs (STORM) analysis revealed FOXO4 family members to be highly enriched upon panobinostat (pano) treatment. Expression levels of FOXO4/1/3 in B) microarray and C) qRT-PCR. D) RT-qPCR analyses revealed a signiﬁ cant increase of FOXO4 upon panobinostat treatment in FOXO4a3 overexpressing (4a3-OE) cells (green) and control Co01 cells (Ctrl-EV) (red). Loss of expression in E) stem cell associated marker LGR5 and increase in expression of F) differentiation marker cytokeratin (KRT20) can be observed in FOXO4 upon panobinostat treatment. G) 4a3-OE cells have innate increased expression of KRT20, which corresponds to distinct morphological changes upon panobinostat treatment and H) loss of clonogenic capacity. I) FOXO4 knockdown in Co01 lines (FOXO4- KD) can be observed upon treatment with doxycycline (DOX) and panobinostat can no longer increase FOXO4 upon knock-down. FOXO4-KD does not affect expression of J) stem cell associate marker LGR5 but it does prevent induction of K) differentiation marker KRT20 upon panobinostat treatment. L) Morphological changes induced by panobinostat are further diminished upon DOX treatment in FOXO4- KD cells. n≥3. not signiﬁ cant (ns). *p value ≤ 0.05 (t-test). **p value ≤ 0.01 (t-test). ***p value ≤ 0.001 (t-test).

152

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 152152 008/05/168/05/16 22:0522:05 Chapter 8

Treatment of the transduced cultures with 100ng/ml doxycycline induced FOXO4 levels to more than 100-fold within hours accompanied by distinct morphological changes suggestive of differentiation (supplementary fi gure 3). Importantly, baseline expression of FOXO4 in these lines was already elevated signifi cantly, pointing to a slight leakiness, which we could exploit to study the effect on colon-CSCs (fi gure 3d). Intriguingly, FOXO family members have previously been implicated to play a tumor suppressor role, and increased cell death upon overexpression has been reported [21]. Correspondingly, single cell-cloned FOXO4a3-OE lines showed elevated levels of background cell death when compared to control cells and upon doxycycline expression cell death was further enhanced (supplementary fi gure 3). In line with a role for FOXO4 in HDACi-induced differentiation, we observed a signifi cant decrease of LGR5 expression levels and increased KRT20 levels in the FOXO4a3-OE cells (fi gure 3e,f). Furthermore, panobinostat could enhance these effects leading to a further reduction in LGR5 and induction of FOXO4 and KRT20 expression levels (fi gure 3d,e,f). Interestingly, despite a signifi cant change in the expression of differentiation markers by FOXO4, the formation of polarized structures in Matrigel could only be observed upon panobinostat treatment (fi gure 3g), suggesting that additional changes next to FOXO4 activation are required. Nevertheless, consistent with the observed decrease in stem cell marker expression FOXO4 did affect stem cell functionality, as spheroid cultures of FOXO4a3-OE cells show a signifi cant loss in their clonogenic capacity (fi gure 3h), which becomes undetectably low when FOXO4 is induced further with doxycycline (results not shown). Over-expression of FOXO4 thus appears to mimic and amplify the effect of panobinostat, suggesting that the effects of panobinostat on colon-CSCs are in part dependent on FOXO4. 8 In order to establish the extent of FOXO4 regulation after panobinostat treatment we generated inducible FOXO4 knockdown (FOXO4-KD) lines, using primary Co01 spheroid cultures. Using these lines, FOXO4 expression could be signifi cantly decreased upon doxycycline stimulation and the induction of FOXO4 that was observed upon panobinostat treatment was effectively prevented (fi gure 3i). Interestingly, the dampening effect of panobinostat on LGR5 expression was retained in FOXO4 knockdown cells (fi gure 3j). This suggests that FOXO4 over-expression can repress LGR5, but that loss of FOXO4 is not suffi cient to prevent HDACi induced repression and likely involves alternative factors as well. In contrast, the HDACi induced expression of differentiation marker KRT20 was fully dependent on FOXO4 and no longer observed in the FOXO4-KD line. In addition, morphological changes were no longer detected when FOXO4 was deleted (fi gure 3k,l). FOXO4-KD hereby seems to effectively block the differentiation effect of panobinostat by preventing increased expression of KRT20 and morphological changes.

153

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 153153 008/05/168/05/16 22:0622:06 Chapter 8

FOXO4 expression controls apoptotic priming via Bcl-XL

FOXO family members have long been implicated to play a tumor suppressive role in cancer, and upon activation inducing cell-cycle arrest can, amongst others, lead to cell senescence, apoptosis or differentiation [22]. Intriguingly, activation of FOXO4 has been implicated to mediate apoptosis by regulating the levels of BCL-XL [23], suggesting that panobinostat induced FOXO4 could potentially regulate the apoptotic threshold via BCL-XL regulation. Interestingly, we have previously reported that colon-CSCs depend BCL-XL in order to establish an anti-apoptotic threshold, preventing chemotherapy-induced apoptosis [4]. Indeed, overexpression of BCL-XL could prevent chemosensitization, whereas specifi c inhibition of BCL-XL could sensitize colon- CSCs to chemotherapeutics. In order to investigate whether the transcriptional program induced by FOXO4 upon panobinostat treatment was regulating the apoptotic threshold via BCL-XL in colon-CSCs we fi rst validated whether the sensitization effect was indeed BCL-XL dependent. Treatment with HDAC inhibitor, panobinostat, signifi cantly reduced expression levels of BCL- XL (fi gure 4a). In order to analyze the effects panobinostat on chemosensitivity in colon-CSCs, we generated a BCL-XL overexpression system. Interestingly, the induced chemosensitivity to oxaliplatin upon panobinostat treatment could be overcome upon BCL-XL overexpression (fi gure 4b), demonstrating the importance of the mitochondrial apoptotic machinery, in particular BCL-XL, for the observed effects. Previously we have shown that differentiated cells have increased sensitivity to ABT-737 than colon-CSCs [4].

In order to demonstrate that panobinostat does modify the apoptotic threshold via modifi cation 8 of the BCL-XL we treated cells with and without panobinostat along a concentration curve of Bcl-2 family members inhibitor, ABT-737 (fi gure 4c). In this manner we could show that IC50 to ABT-737 in colon-CSCs as well as in differentiated cells drops signifi cantly upon treatment with panobinostat (fi gure 4d), demonstrating the ability of HDAC inhibitors to exploit dependency on BCL-XL and lower the apoptotic threshold helping to facilitate chemotherapy. The effect of HDAC inhibition is dependent on transcriptional activation and we show a signifi cant enrichment of FOXO4 transcriptional targets. FOXO4 has previously been implicated to negatively regulate BCL-XL expression [23]; therefore we continued to investigate the effects of FOXO4 on BCL-XL regulation and chemosensitization. Overexpression of FOXO4 in FOXO4a3-OE cells demonstrated decreased expression levels of BCL-XL, which was further diminished upon treatment with panobinostat (fi gure 4e). As previously reported [21] increased levels of background apoptosis, measured by activated caspase-3, could be observed in FOXO4a3-OE cells. Furthermore, FOXO4a3-OE colon-CSCs were highly sensitive to cell death in the presence of oxaliplatin alone (fi gure 4f), suggesting that the increased levels of FOXO4 are enough to sensitize the colon-CSCs to cell death. In addition the FOXO4-KD cells were no longer sensitive to panobinostat and showed no down-regulation of BCL-XL upon panobinostat treatment (fi gure 4g). Furthermore, chemosensitization of colon-CSCs could

154

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 154154 008/05/168/05/16 22:0622:06 Chapter 8

not be induced upon FOXO4 knockdown (ﬁ gure 4h), suggesting that the chemosensitization effects observed upon panobinostat treatment in colon-CSCs are indeed driven by FOXO4 transcriptional regulation.HDAC inhibitors hereby induce differentiation and sensitize colon- CSCs to conventional chemotherapeutic. By exploiting the colon-CSC dependency on BCL-XL HDAC inhibitors thus lower the apoptotic threshold helping to facilitate chemotherapy, in a FOXO4 dependent manner (ﬁ gure 5).

A.B. C. Diff CSC D. BCL-XL Diff + Pano CSC + Pano 5 100 ** 40 3000 Ctrl-EV * * * BCL-XL-OE 4 80 30 2000 3 60

20 2 40 IC50 (nM) 1000 10 1 20 % Caspase 3 active cells % Caspase 3 active cells Expression (relative to B2M) 0 0 0 0 Pano - + Pano ---++ -++ 10 100 1000 10000 Pano - + - + Oxali -++- -++- ABT-737 (nM) Diff CSC

E. F. G. H. BCL-XL BCL-XL 60 ** ** ** ** ** 40 4 ** ** * * 1.0 30 3 40

20 2 0.5 20 % AnnexinV +ve Cells 1 10 % Caspase 3 active cells Fold change in Expression Expression (relative to B2M) 0 0 0 0 Pano - + - + Pano - + - + - + -+ Pano-+ + - Pano - + - + - + -+ Oxali - - + + - - + + DOX-+ - + Oxali - - + + - - + + DOX --- - ++++ 8 Ctrl-EV 4a3-OE Ctrl-EV 4a3-OE shFOXO4 shFOXO4

Figure 4: Bcl-XL regulation is required for chemosensitization effect of panobinostat in colon-CSCs. A) qRT-PCR analysis shows BCL-XL levels drop after treatment with panobinostat (pano). B) Flow cytometry analysis of activated Caspase-3 shows that overexpression of BCL-XL (BCL-XL-OE) overcomes the sensitization effect of panbinostat upon oxaliplatin treatment observed in control cells (Ctrl-EV). C) Inhibition of BCL-XL using ABT-737 at variable concentrations shows different sensitivities, based on ﬂ ow cytomotery caspase-3 activity in red the colon-CSCs (Red) and differentiated cells (Blue) treated with and without panobinostat. D) The corresponding IC-50 values of cells treated with ABT737 show increased sensitization upon BCL-XL inhibition by ABT-737 in both differentiated (blue) and CSCs (Red). E) qRT-PCR analysis of FOXO4a3-OE cells corresponds with decreased BCL-XL expression. F) Activated Caspase-3 measured by ﬂ ow cytometry shows FOXO4a3-OE cells to be highly sensitized to treatment with oxaliplatin. G) Upon knockdown of FOXO4 (shFOXO4) with doxycycline (DOX) stimulation, Bcl- XL is no longer regulated upon panobinostat treatment and H) AnnexinV staining reveals a decreased sensitization of in CSCs, DOX stimulation in treatments with oxaliplatin and panobinostat. n≥3. *p value ≤ 0.05 (t-test). **p value ≤ 0.01 (t-test). ***p value ≤ 0.001 (t-test)

155

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 155155 008/05/168/05/16 22:0622:06 Chapter 8

Discussion

In this study we used a small-scale drug screen and identifi ed HDAC inhibitors as candidates to sensitize colon-CSCs to oxaliplatin. By the removal of acetyl groups from lysine residues on proteins HDACs are able to change the function, stability and activity of proteins, having a signifi cant impact on transcriptional regulation in a cell [24]. In agreement, when blocking de novo protein transcription and translation we show that HDACi-induced sensitization is dependent on a change in transcriptional regulation. Moreover, this coincides with a signifi cant change in the morphology of spheroids over time, which is consistent with an increase in differentiation and loss of stemness. In addition, our data provides evidence for an HDACi-induced decrease in BCL-XL and as a consequence enhanced chemotherapy sensitivity. These effects are driven to a large extent by an enrichment of FOXO4 transcriptional activity, as over-expression of FOXO4 mimics and enhances the impact of HDACi treatment, whereas, the knockdown of FOXO4 blocks HDACi-mediated differentiation and BCL-XL suppression.

Differentiated Cells † Differentiated Cells

Chemotherapeutic (eg: Oxaliplatin)

HDACi FOXO4 BCL-XL 8 Wnt Chemotherapeutic (eg: Oxaliplatin) Stem Cells Stem Cell survival

Figure 5: Model depicting chemosensitization of colon-CSCs upon treatment with HDACi. Upon HDACi treatment FOXO4 is increased, decreasing the levels of BCL-XL

Previously we have demonstrated that colon-CSCs are dependent on BCL-XL to overcome therapeutic insults [4]. This apoptotic threshold has important implications for the effi cacy of chemotherapy as recent data indicate that the priming state of mitochondria in cancer cells, which can be measured ex-patient using BH3 profi ling, directly relates to the outcome of therapy both in hematopoietic malignancies as well as in ovarian cancer [25]. This proposes that modulating the apoptotic threshold in colon-CSCs using direct intervention with BH3 mimetics [4] or, alternatively, with compounds that affect the expression levels of key players that regulate this threshold is an intriguing new therapeutic avenue. Here we now provide compelling evidence to show that HDAC inhibitors alter BCL-XL levels, lowering the apoptotic threshold in colon- CSCs as well as in more differentiated cells, involving a FOXO4-dependent modulation (fi gure 5).

156

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 156156 008/05/168/05/16 22:0622:06 Chapter 8

FOXO transcription factors have been intimately linked to the regulation of cell death [26-28] and strikingly this capacity has been directly coupled to the acetylation status of for instance FOXO3a [29], where acetylated FOXO would drive pro-apoptotic signaling through BIM and FASL expression [26, 30]. FOXO acetylation could thus be a means to modulate cell death and explain the observed sensitization. However, FOXO acetylation is regulated by sirtuins, which are insensitive to the HDAC inhibitors used in this study, excluding a direct link. More importantly, we do not observe a change in acetylation status for either FOXO4 or FOXO3a (data not shown), suggesting that this does not explain the observed effects of HDAC inhibition on pro-apoptotic signaling. In addition, BIM expression is not changed in colon-CSCs upon HDACi treatment. We therefore favor a model in which BCL-XL expression is dampened by FOXO4 facilitating chemotherapy-induced apoptosis. Importantly, FOXO4 has previously been reported to regulate BCL-XL [23] through a BCL-6-dependent mechanism, conﬁ rming the potential of FOXO4 to directly regulate the apoptotic threshold.

A direct role for FOXO4 in the regulation of cancer stemness appears to exist as well, although this may be more complex. Over-expression of FOXO4 blocks cancer stemness and reduces LGR5 expression, consistent with the observation that HDACi enhance FOXO4 and reduce stemness. However, FOXO4 knockdown does not prevent the HDACi-induced reduction in LGR5. As LGR5 is reportedly a canonical WNT target gene regulated by beta-catenin/ TCF transcription factors [31], this suggests that HDAC inhibitors could affect canonical WNT signaling in additional ways, which may involve acetylation of beta-catenin itself [32, 33]. Importantly, the LGR5 regulation by FOXO4 points to another striking observation as it conﬂ icts with earlier observations in which we reported that LGR5 was induced by over- 8 expression of FOXO3a3 in DLD1 cells [34]. This is not related to culture conditions or cell line differences as over-expression of FOXO3a3 in our colon-CSCs also resulted in LGR5 induction (data not shown). Differential regulation of LGR5 by FOXO4 and FOXO3a3 are in in line with the observed HDACi-mediated induction of FOXO4 but not FOXO3a expression. However, what determines this differential regulation of LGR5 by FOXOs remains currently unclear. Importantly, a pro-tumorigenic role for FOXO3a in CRC was also reported by Tenbaum et al., who proposed a model in which ß-catenin and FOXO3a cooperate to drive tumor metastasis [35]. Whether this relates to the currently observed differences is not clear. We do believe, however, our current observations are in line with the observed association between FOXO4 and ß-catenin, where we showed that FOXO4 can compete with ß-catenin for TCF/LEF interaction [36, 37]. Over-expression of FOXO4 could hereby repress canonical WNT pathway activation and thus LGR5 expression. Importantly, WNT pathway activity is directly linked to cancer stemness [12] and as such could explain the observed repression exerted by HDACi on stemness.

How then is FOXO4 activated by HDAC inhibition? Although our data do not provide conclusive

157

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 157157 008/05/168/05/16 22:0622:06 Chapter 8

evidence for the exact regulation, we believe that the enhanced expression of FOXO4 is a likely explanation of the observed effects. Elevated expression of FOXO4 is very rapid (within hours) after addition of HDACi and precedes the observed induction of FOXO1 (data not shown). This suggests that increased FOXO4 expression is the fi rst event, which can subsequently lead to enhanced expression of FOXO1, which is consistent with the induction of FOXO1 observed upon FOXO4a3 over-expression and the reported regulation of FOXO transcription factors by the different family members [38]. Intriguingly, both HDAC1 and HDAC2 have been reported to regulate FOXO1 and FOXO3a, potentially through a direct association and/or recruitment to specifi c promoter regions [39]. Whether this also holds for FOXO4 remains to be established but is an interesting possibility, especially as HDAC2 has been implicated in CRC development. For instance, increased expression of HDAC2 in CRC lesions has previously been reported [40] and APCmin mice have increased expression of HDAC2 within the adenoma lesions. Moreover, treatment with the HDAC inhibitor valproic acid reduces polyp formation and size [40], whereas APCmin-driven transformation in an HDAC2 knockout background is strongly diminished, confi rming the necessity for HDAC2 to allow tumor formation to occur [41].

Our data thus point to a potential combination therapy involving HDACi-induced sensitization and differentiation with chemotherapy-induced cell death. Importantly, several HDAC inhibitors have been FDA approved for the use of hematological malignancies and clinical trials for solid malignancies are currently ongoing. However, initial study results have been disappointing, with almost no response to treatment even in combination therapies [42, 43]. Moreover, several trials were prematurely terminated, due to toxic side effects including severe diarrhea [42, 44]. The 8 latter could be related to the role we recently unraveled for both HDAC1 and HDAC2 in the maintenance of normal intestinal stem cells [45]. Similarly, hematopoietic toxicities of HDACi could be related to the role of HDAC1 and HDAC2 in the hematopoietic stem cell [46]. Next to the fact that these mouse models provide a rational to toxicities observed in the clinic, they also provide an insight into the therapies that may avoid these toxicities. In both settings deletion of HDAC1 in combination with HDAC2 was required to delete stemness, suggesting that targeted inhibition of HDAC1 or HDAC2 individually would leave normal stem cells untouched. Whether this would affect the sensitivity of colon-CSCs remains to be determined, but the observed induction of HDAC2 in CRC could point to a window-of opportunity. Alternatively, our data could provide a rational for combining HDACi with BH3 mimetics targeting BCL-XL. Clinical targeting of BCL-XL has proven to be relatively toxic due to the dependency of thrombocytes on BCL-XL. However, as we show HDACi’s to lower the IC50 of the cells by almost 10-fold (ﬁ gure 4d), a combinatory treatment would potentially decrease the required dosage and allow its usage in patients. Furthermore, normal intestinal stem cells appear insensitive to the actions of BH3 mimetics this may provide an interesting combination therapy. In summary, we identiﬁ ed a novel pathway for the effect of HDAC inhibitors, where the induction of FOXO4

158

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 158158 008/05/168/05/16 22:0622:06 Chapter 8

induces differentiation and down-regulates BCL-XL expression, forcing colon-CSCs into a non- clonogenic state and enhancing sensitivity to chemotherapeutics.

Materials and Methods

Colon Spheroid cultures and plasmid construction

Spheroid cultures were derived from primary tumors (Co) or liver metastasis (LM) of colorectal cancer patients and maintained in ultra-low adherent fl asks (Corning). Isolation methods are described in [47]. Cells were grown in advanced DMEM/F12 (Gibco) supplemented with N2 (Gibco), 2 mM L-glutamine, 6 mg/ml glucose, 5 mM HEPES, 4 μg/ml heparin, 50 ng/ ml epidermal growth factor (EGF) and 10 ng/ml basic fi broblast growth factor (bFGF). To generate transduced cultures, cells were lentiviral transduced and single-cell sorted in 96-well ultra-low adhesion plates (Corning) with FACSaria (BD Biosciences). The TOP-GFP vector was a gift from Dr. Laurie Ailles and was described previously [48]. Ectopically BCL-XL overexpressing cells were generated by transduction with pHEFTIR-BCL-XL as described before [4]. Knockdown of FOXO4 was performed with the pTRIPZ shRNA lentiviral plasmid Clone ID: V3THS_358494 (Dharmacon – open Biosystems). The overexpressing FOXO4a3 lines were generated using a modifi ed pTRIPZ vector. The RFP was removed using AgeI and HpaI restriction sites. Linearized DNA fragments containing additional restriction sites EcoRI and BstXI were added between AgeI and HpaI and the vector was sealed, Using restriction sites EcoRI and XhoI the vector was opened up and large dsDNA DNA fragments (GeneArt® Strings™ DNA Fragments, Invitrogen) were inserted using SLIC cloning [49]. The dsDNA 8 segments FOXO4a3 were designed with a HA-tag at the N-terminal and mutations to alanine at Thr32, Ser187 and Ser252.

Cell death screening assays

50 000 TOP-GFP transduced colon-CSCs were seeded in a 12-wells culture plate (Greiner, Alphen a/d Rijn, The Netherlands) overnight and allowed to adhere. Next day dead cells were washed off and a cells were pretreated with sensitizing compounds for indicated time periods followed by 24 hr treatment with oxaliplatin. Cells were then collected and measured for apoptotic cell death using staining for activated caspase-3 or Annexin V. Activated Caspase-3 activity was measured in colon-CSCs according to the manufacturer’s instructions (BioVision, Milpitas, CA, USA) and described by [4]. In short, cells were collected with trypsin-EDTA and washed once with medium and stained with RED-DEVD-FMK for 1 h at 37 °C. Subsequently, cells were washed twice with wash buffer and ﬂ ow cytometry was performed with FACS canto (BD biosciences). Cell death was measured in colon-CSCs and differentiated cancer cells respectively by gating on TOP-GFPhigh cells and TOP-GFPlow cells. All sensitization treatments for the drug screening were done with 1 hr drug stimulation before the addition of oxaliplatin. Fold

159

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 159159 008/05/168/05/16 22:0622:06 Chapter 8

induction was measured by the cell death induced in colon-CSCs in the combination treatment compared to oxaliplatin treatment alone in the colon-CSCs. All drugs in the screening were tested in duplicate in two dosages (overview supplementary table 1). Using Annexin V we detected phospatidylserine exposure. Staining was performed according to the manufacturers protocol Annexin V (BD Biosciences). In short, spheroid cultured cells were stained with Annexin V-APC (BD Biosciences) and 7-AAD (BD Biosciences) in Annexin V binding buffer (BD Biosciences) for 15 min at RT before measurement.

Xenograft studies

To generate in vivo tumors, 5000 colon-CSCs (TOP-GFPhigh cells) were isolated by FACS sorting, mixed at an 1:1 ratio with Matrigel and injected subcutaneously into nude (Hsd: Athymic Nude/ Nude) (Harlan) mice. Treatments were started when tumors reached a size between 50 and 100 mm3. Mice were treated intra-peritoneal for 4 consecutive weeks either with daily injection (5 times/week) of 10 mg/kg panobinostat (Selleck Chemicals) dissolved in PBS/2% DMSO, or once a week with 1 mg/kg oxaliplatin (Sigma) dissolved in PBS, or with a combination of both drugs. After 4 weeks treatments were stopped and tumor growth was measured. Mice reaching a tumor size 1000 mm3 were sacriﬁ ced in accordance with ethical regulation in The Netherlands. All mice experiments were performed in accordance with the Animal Experimentation Committee (DEC) at the Academic Medical Center in Amsterdam

Reagents

8 Sensitization assays were conducted with HDAC inhibitor treatment followed by the addition of 24 hr Oxaliplatin (50 μM, Sigma-Aldrich). The following HDAC inhibitors were tested for 16 hr unless otherwise stated: vorinostat (2 μM, Selleck Chemicals), Panobinostat (10 nM, Selleck Chemicals), MS-275 (1 μM, Selleck Chemicals), Sodium Butyrate (8 mM, Sigma Aldrich), Trichostatin A (1 μM, Sigma Aldrich), Valproic acid (1 μM, Sigma Aldrich). 100 ng/ml doxycycline (Sigma-Aldrich) was added for 24 hr to FOXO4a3-OE lines to induce FOXO4 expression and 1000 ng/ml doxycycline was added for 48 hr to shFOXO4 knockdown lines to knockdown FOXO4. Treatments with Actinomycin D (1ug/ml, Sigma-Aldrich) and Cycloheximide (10 ug/ ml, Sigma-Aldrich) were conducted 1 hr prior to HDACi addition. ABT-737 (Selleck Chemicals) was used at varying concentrations and added for 24 hr prior to measurement.

Gene expression proﬁ ling and cluster analysis

At least 100.000 cells were sorted and RNA was isolated with Trizol reagent (Invitrogen) according to the manufacturer’s instructions. For microarray, RNA quality was determined using the RNA 6000 Nano assay on the Agilent 2100 Bioanalyzer (Agilent Technologies). Gene expression proﬁ ling was performed by Micro-Array Department (MAD) (CCG, Cologne) using

160

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 160160 008/05/168/05/16 22:0622:06 Chapter 8

Human Genome U133 Plus 2.0 microarrays in accordance with the manufacturer’s protocol (Affymetrix). For clustering analysis a stem cell signature was used that was published before [18]. First expression of these genes in differentiated tumor cells and panobinostat treated colon- CSCs relative to colon-CSCs treated with DMSO was calculated. Next Pearson correlation, average linkage in the Multi experiment Viewer package 4.5 was used to cluster the samples [50].

Transcription Factor Binding Site Enrichment Analysis

Differentially expressed genes were identifi ed using the R2 platform (http://r2.amc.nl). An ANOVA analysis was used with a p-value cutoff of 0.01. In order for a gene to be considered signifi cantly regulated it also had to have a minimal highest expression value of 50 and a minimum number of 2 ‘present’ calls within the 6 samples compared. Genes signifi cantly upregulated and downregulated by Panobinostat treatment were analyzed separately. The motif analysis was carried out as in [51] with the only procedural exception being the usage of TRANSFAC database version 2013.2 vertebrate matrices. The upregulated genes were used to identify ‘foreground’ promoters (832) while ‘background’ promoters were randomly drawn from genes that were not signifi cantly regulated (1527) in the experiment (p-value > 0.01).

Matrigel differentiation assay and immunohistochemistry

2000 single cells were mixed with 50 μl 50% Matrigel (BD Bioscience) and seeded as a droplet in a pre-heated 12 wells plate (Greiner). After 10 min incubation at 37 °C, 1 ml medium was added. After 3 days medium was refreshed with medium containing HDAC inhibitors or DMSO. After 4 days of treatment cells were fi xed with 4 % paraformaldehyde overnight, dehydrated 8 in a standard ethanol/xylene series and embedded in paraffi n. 5μM sections were stained with 1:200 mouse monoclonal antibody against CK20 (clone GTX15205, GeneTex), using standard procedures according to the manufacturers protocol. Quantifi cation of the polarized structures were quantifi ed blindly by three different individuals.

Flow cytometry

Staining for stem cell markers CD133 and LGR5 was performed in colon-CSCs (GFPhigh population). After 16 h of HDACi treatment, cells were dissociated and stained as described before [52] with AC133/CD133-APC antibody (1:25, Miltenyi Biotec) or anti-LGR5-biotin antibody (4D11F8, 1:100, BD Biosciences). For LGR5 staining, cells were incubated with APC conjugated streptavidin (1:500, E-biosciences) and washed twice with PBS containing 1% bovine serum albumin. Dead cells were excluded with 7-AAD (BD Biosciences) and stainings were analyzed on a FASCanto (BD Biosciences).

161

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 161161 008/05/168/05/16 22:0622:06 Chapter 8

RNA extraction, cDNA synthesis, qRT-PCR

Total RNA was extracted according to manufacturer’s instructions using the NucleoSpin RNA kit (Macherey-Nagel). RNA quantity and quality were assessed using Thermo- Scientifi c nanodrop 1000 Spectrophotometer (Thermo-Fischer Scientifi c). Following DNAse treatment using DNAseI Amplifi cation Grade (Life Technologies), cDNA was synthesized using SuperScript III reverse transcriptase (Life Technologies) with random primers (Life Technologies). Quantitative RT-PCR was performed using 2x SYBR Green Master Mix (Roche), and transcript levels were normalized to that of ß-2-microglobulin (B2M). Primers used in this study were as follows B2M: forward, 5’-CTTCAGTCGTCAGCATGG- 3’; reverse, 5’-GTTCTTCAGCATTTGGATTTC-3’. LGR5: forward: 5’- AATCCCCTGCCCAGTCTC-3’, reverse: 5’- CCCTTGGGAATGTATGTCAGA-3’. KRT20: forward: 5’- TGTCCTGCAAATTGATAATGCT-3’, reverse 5’- AGACGTATTCCTCTCTCACTCTCATA -3’. FOXO1: forward, 5’- CAAGAGCGTGCCCTACTTCA-3’, reverse, 5’- CTTGCCACCCTCTGGATTGA -3’. FOXO3: forward, 5’- CTTCAAGGATAAGGGCGACA -3’, reverse, 5’- AGTTCCCTCATTCTGGACCC-3’. FOXO4, forward 5’- GAGGCTCCCGCCGGAATG -3’. reverse, 5’- GGTTGTGGCGGATCGAGTT-3’. BCL-XL: forward 5’- AGAGAACAGGACTGAGGCCC-3’, reverse 5’- TCAAAGCTCTGATATGCTGTCCC-3’.

Limiting-dilution assay

8 In order to quantify the clonogenic capacity present in the 3D-primary spheroid cultures a limiting dilution assay was performed. Colon-CSCs were seeded adherently and treated accordingly for 16 hrs panobinostat and if required 48hrs doxycycline. After dissociation colon-CSCs (10% TOP-GFPhigh) were sorted with FACSaria (BD Biosciences) and were deposited into ultra- low adherence 96-well plates (Corning) in a limiting dilution fashion at 1, 2, 4, 8, 16, 24, 32, 64 and 128 cells per well. After two weeks clonogenic capacity was evaluated by scoring wells with spheres and calculated using the Extreme Limiting Dilution Analysis ‘limdil’ function as described [53].

Western Blot

The cells were lysed in Cell Lysis buffer (Cell Signaling Technology) containing protease/ phosphatase inhibitor cocktail (Cell Signaling Technology). Protein quantiﬁ cation was done using the Pierce BCA Protein Assay Kit, and 20ug was loaded per well of 4–15% precast gels (Bio-Rad), transferred to a Hybond-P membrane (GE Healthcare Life Sciences), and blocked for 1 hour in 5% bovine serum albumin in Tris-buffered saline and Tween 20 (0.1%). Primary antibodies for overnight 4°C incubation were 1:1000 FOXO 4 (PA5-28927, Thermo Scientiﬁ cs), PARP (65196E, BD Bioscience). After three 10 min washes with TBS-T (0.1%), the secondary

162

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 162162 008/05/168/05/16 22:0622:06 Chapter 8

antibody 1:2000, anti-rabbit-horseradish peroxidase (#7074; Cell Signaling Technology) or 1:5000 anti-mouse (#1070-05, Southern Biotech) was applied for 1 hour at room temperature. After washing, the membrane was developed using the ImageQuanti LAS4000 (GE Healthcare Life Sciences). Protein was normalized on α-tubulin (H235; Santa Cruz).

Acknowledgements

We would like to thank B. Hooibrink and T. van Capel for assistance with ﬂ uorescence-activated cell sorting experiments.

Conﬂ ict of interest

The authors declare no conﬂ ict of interest

163

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 163163 008/05/168/05/16 22:0622:06 Chapter 8

References

1. Howlader N, N.A., Krapcho M, Garshell 9. Lombardo, Y., et al., Bone morphogenetic J, Miller D, Altekruse SF, Kosary CL, Yu protein 4 induces differentiation of colorectal M, Ruhl J, Tatalovich Z,Mariotto A, Lewis cancer stem cells and increases their response to DR, Chen HS, Feuer EJ, Cronin KA chemotherapy in mice. Gastroenterology, (eds). SEER Cancer Statistics Review, 1975- 2011. 140(1): p. 297-309. 2012. based on November 2014 SEER 10. Fearon, E.R., Molecular genetics of colorectal data submission, posted to the SEER web cancer. Annu Rev Pathol, 2011. 6: p. 479- site]. Available from: http://seer.cancer. 507. gov/csr/1975_2012/. 11. Fearon, E.R. and B. Vogelstein, A genetic 2. Van Emburgh, B.O., et al., Acquired model for colorectal tumorigenesis. Cell, 1990. resistance to EGFR-targeted therapies in 61(5): p. 759-67. colorectal cancer. Mol Oncol, 2014. 8(6): p. 12. Vermeulen, L., et al., Wnt activity deﬁ nes 1084-94. colon cancer stem cells and is regulated by the 3. Medema, J.P., Cancer stem cells: the challenges microenvironment. Nat Cell Biol, 2010. 12(5): ahead. Nat Cell Biol, 2013. 15(4): p. 338-44. p. 468-76. 4. Colak, S., et al., Decreased mitochondrial 13. Prasetyanti, P.R., et al., Regulation of stem priming determines chemoresistance of colon cell self-renewal and differentiation by Wnt and cancer stem cells. Cell Death Differ, 2014. Notch are conserved throughout the adenoma- 21(7): p. 1170-7. carcinoma sequence in the colon. Mol Cancer, 8 5. Todaro, M., et al., Colon cancer stem cells 2013. 12(1): p. 126. dictate tumor growth and resist cell death by 14. Voloshanenko, O., et al., Wnt secretion is production of interleukin-4. Cell Stem Cell, required to maintain high levels of Wnt activity 2007. 1(4): p. 389-402. in colon cancer cells. Nat Commun, 2013. 4: 6. Dylla, S.J., et al., Colorectal cancer stem cells p. 2610. are enriched in xenogeneic tumors following 15. Colak, S. and J.P. Medema, Cancer stem cells- chemotherapy. PLoS One, 2008. 3(6): p. -important players in tumor therapy resistance. e2428. FEBS J, 2014. 281(21): p. 4779-91. 7. Hoey, T., et al., DLL4 blockade inhibits 16. Roth, S.Y., J.M. Denu, and C.D. Allis, tumor growth and reduces tumor-initiating cell Histone acetyltransferases. Annu Rev frequency. Cell Stem Cell, 2009. 5(2): p. 168- Biochem, 2001. 70: p. 81-120. 77. 17. Thiagalingam, S., et al., Histone deacetylases: 8. Wielenga, M.C., et al., ER-Stress-Induced unique players in shaping the epigenetic histone Differentiation Sensitizes Colon Cancer Stem code. Ann N Y Acad Sci, 2003. 983: p. 84- Cells to Chemotherapy. Cell Rep, 2015. 13(3): 100. p. 490-4.

164

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 164164 008/05/168/05/16 22:0622:06 Chapter 8

18. de Sousa, E.M.F., et al., Methylation of cancer- 27. Stahl, M., et al., The forkhead transcription stem-cell-associated Wnt target genes predicts poor factor FoxO regulates transcription of p27Kip1 prognosis in colorectal cancer patients. Cell Stem and Bim in response to IL-2. J Immunol, Cell, 2011. 9(5): p. 476-85. 2002. 168(10): p. 5024-31. 19. Schones, D.E., A.D. Smith, and M.Q. 28. Dijkers, P.F., et al., Expression of the pro- Zhang, Statistical signifi cance of cis-regulatory apoptotic Bcl-2 family member Bim is regulated modules. BMC Bioinformatics, 2007. 8: p. by the forkhead transcription factor FKHR-L1. 19. Curr Biol, 2000. 10(19): p. 1201-4. 20. Essaghir, A., et al., The transcription of 29. Brunet, A., et al., Stress-dependent regulation FOXO genes is stimulated by FOXO3 and of FOXO transcription factors by the SIRT1 repressed by growth factors. J Biol Chem, deacetylase. Science, 2004. 303(5666): p. 2009. 284(16): p. 10334-42. 2011-5. 21. Zhang, X., et al., Akt, FoxO and regulation 30. Dijkers, P.F., et al., FKHR-L1 can act as a of apoptosis. Biochim Biophys Acta, 2011. critical effector of cell death induced by cytokine 1813(11): p. 1978-86. withdrawal: protein kinase B-enhanced cell 22. Ho, K.K., S.S. Myatt, and E.W. Lam, survival through maintenance of mitochondrial Many forks in the path: cycling with FoxO. integrity. J Cell Biol, 2002. 156(3): p. 531-42. Oncogene, 2008. 27(16): p. 2300-11. 31. Barker, N., et al., Identifi cation of stem cells in 23. Tang, T.T., et al., The forkhead transcription small intestine and colon by marker gene Lgr5. factor AFX activates apoptosis by induction Nature, 2007. 449(7165): p. 1003-7. of the BCL-6 transcriptional repressor. J Biol 32. Wolf, D., et al., Acetylation of beta-catenin by 8 Chem, 2002. 277(16): p. 14255-65. CREB-binding protein (CBP). J Biol Chem, 24. Haberland, M., R.L. Montgomery, and 2002. 277(28): p. 25562-7. E.N. Olson, The many roles of histone 33. Levy, L., et al., Acetylation of beta-catenin deacetylases in development and physiology: by p300 regulates beta-catenin-Tcf4 interaction. implications for disease and therapy. Nat Rev Mol Cell Biol, 2004. 24(8): p. 3404-14. Genet, 2009. 10(1): p. 32-42. 34. Eijkelenboom, A., et al., Genome-wide 25. Montero, J., et al., Drug-induced death analysis of FOXO3 mediated transcription signaling strategy rapidly predicts cancer response regulation through RNA polymerase II profi ling. to chemotherapy. Cell, 2015. 160(5): p. 977- Mol Syst Biol, 2013. 9: p. 638. 89. 35. Tenbaum, S.P., et al., beta-catenin confers 26. Brunet, A., et al., Akt promotes cell survival resistance to PI3K and AKT inhibitors and by phosphorylating and inhibiting a Forkhead subverts FOXO3a to promote metastasis in transcription factor. Cell, 1999. 96(6): p. 857- colon cancer. Nat Med, 2012. 18(6): p. 892- 68. 901.

165

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 165165 008/05/168/05/16 22:0622:06 Chapter 8

36. Hoogeboom, D., et al., Interaction of FOXO 46. Bartels, M., et al., Histone deacetylase with beta-catenin inhibits beta-catenin/T cell inhibition modulates cell fate decisions during factor activity. J Biol Chem, 2008. 283(14): myeloid differentiation. Haematologica, 2010. p. 9224-30. 95(7): p. 1052-60. 37. Essers, M.A., et al., Functional interaction 47. Prasetyanti, P.R., et al., Isolation and between beta-catenin and FOXO in oxidative propagation of colon cancer stem cells. Methods stress signaling. Science, 2005. 308(5725): p. Mol Biol, 2013. 1035: p. 247-59. 1181-4. 48. Reya, T., et al., A role for Wnt signalling 38. Calnan, D.R. and A. Brunet, The FoxO code. in self-renewal of haematopoietic stem cells. Oncogene, 2008. 27(16): p. 2276-88. Nature, 2003. 423(6938): p. 409-14. 39. Beharry, A.W., et al., HDAC1 activates FoxO 49. Jeong, J.Y., et al., One-step sequence- and and is both sufﬁ cient and required for skeletal ligation-independent cloning as a rapid and muscle atrophy. J Cell Sci, 2014. 127(Pt 7): p. versatile cloning method for functional genomics 1441-53. studies. Appl Environ Microbiol, 2012. 40. Zhu, P., et al., Induction of HDAC2 78(15): p. 5440-3. expression upon loss of APC in colorectal 50. Saeed, A.I., et al., TM4: a free, open-source tumorigenesis. Cancer Cell, 2004. 5(5): p. system for microarray data management and 455-63. analysis. Biotechniques, 2003. 34(2): p. 41. Zimmermann, S., et al., Reduced body size 374-8. and decreased intestinal tumor rates in HDAC2- 51. Santo, E.E., et al., FOXO3a is a major target 8 mutant mice. Cancer Res, 2007. 67(19): p. of inactivation by PI3K/AKT signaling in 9047-54. aggressive neuroblastoma. Cancer Res, 2013. 42. West, A.C. and R.W. Johnstone, New and 73(7): p. 2189-98. emerging HDAC inhibitors for cancer treatment. 52. Colak, S. and J.P. Medema, Human colonic J Clin Invest, 2014. 124(1): p. 30-9. ﬁ broblasts regulate stemness and chemotherapy 43. Nebbioso, A., et al., Trials with ‘epigenetic’ resistance of colon cancer stem cells. Cell Cycle, drugs: an update. Mol Oncol, 2012. 6(6): p. 2014: p. 0. 657-82. 53. Hu, Y. and G.K. Smyth, ELDA: extreme 44. Slingerland, M., H.J. Guchelaar, and H. limiting dilution analysis for comparing depleted Gelderblom, Histone deacetylase inhibitors: an and enriched populations in stem cell and other overview of the clinical studies in solid tumors. assays. J Immunol Methods, 2009. 347(1- Anticancer Drugs, 2014. 25(2): p. 140-9. 2): p. 70-8. 45. Zimberlin, C.D., et al., HDAC1 and HDAC2 collectively regulate intestinal stem cell homeostasis. FASEB J, 2015. 29(5): p. 2070- 80.

166

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 166166 008/05/168/05/16 22:0622:06 Chapter 8

A) Ctrl 20nM 50nM 100nM

Supplementary Figure 1: Panobinostat stimulation induces distinct morphological changes in primary Co01 spheriod cells at 20nM, 50nM and 100nM

167

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 167167 008/05/168/05/16 22:0622:06 Chapter 8

A) Co108 LGR5 KRT20 3 * 2.5 100 * 2.0 80

2 1.5 60

1.0 40 1

0.5 20 % non-polarized structures Expression (relative to B2M) Expression (relative to B2M) 0 0 Pano- + Pano- + 0 Ctrl FCS B) LM23 Pano LGR5 KRT20 1.0 3 100 * ** 0.8 80

2 0.6 60

0.4 40 1

0.2 20 % non-polarized structures Expression (relative to B2M) Expression (relative to B2M) 0 0 0 Pano- + Pano- + Ctrl FCS Pano C) Co123

2.0 LGR5 0.008 KRT20 100

80 1.5 0.006

60 1.0 0.004 40

0.5 0.002 20 % non-polarized structures Expression (relative to B2M) Expression (relative to B2M) 8 0 0 0 Pano- + Pano- + Ctrl FCS D) Co09 Pano

LGR5 KRT20 0.8 0.010 *

* 0.008 0.6

0.006 0.4 0.004

0.2 0.002 Expression (relative to B2M) Expression (relative to B2M) 0 0 Pano- + Pano- +

Supplementary Figure 2: The effect of panobinostat on multiple different primary CRC spheroid cultures A) Co108, B) LM23 a culture isolated from a liver metastasis, C) Co123 and D) Co09. qRT- PCR shows the effect of panobinostat on stemness marker LGR5 and differentiation marker cytokeratin 20 (KRT20). On the right is a quantification of the polarized structures observed upon panobinostat treatment, when cells are embedded in matrigel.. n 3. *p value 0.05 (t-test). **p value 0.01 (t-test). ***p value 0.001 (t-test)

168

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 168168 008/05/168/05/16 22:0622:06 Chapter 8

B. A. FOXO4 FOXO1 FOXO3 4 0.20 0.0010 4a3-OE Cyto Nucl 4a3-OE + DOX 4hrs *** ** - + - + ** * 4a3-OE + DOX 16hrs Pano ** ** 0.0008 4a3-OE + DOX 24hrs 3 0.15

0.0006 FOXO4 55kD 2 0.10 0.0004 PARP 116kD 1 0.05 0.0002 Expression (relative to B2M) Expression (relative to B2M) Expression (relative to B2M)

0 0 0 C. + DOX D. 30 *** ** ***

* ** 10 % Cell Death (7AAD+ve)

Ctrl 0 Ctrl 4a3-OE Ctrl + DOX 4a3-OE+DOX

4a3-OE x20 8

Supplementary Figure 3: A) FOXO4 protein expression can only be detected by immunoblotting in the nucleus and not in the cytoplasm B) Overexpressing FOXO4a3 (FOXO4a3-OE) cells can be induced by doxycycline (DOX) stimulation, DOX induces expression levels of FOXO4 and FOXO1 but not FOXO3. C) DOX stimulation induces distinct morphological changes in FOXO4a3-OE cells and D) corresponding with increased background cell death shown by PI positivity. n 3. *p value 0.05 (t-test). **p value 0.01 (t-test). ***p value 0.001 (t-test)

169

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 169169 008/05/168/05/16 22:0622:06 Chapter 8

Supplementary Table 1: Full list and results of screening compounds used in the small-scale drug screen

signaling compound [uM] FC [uM] FC 1 Erlotinib 0,1 1,2 1 1,5 2 Desatinib 0,1 1,4 1 0,8 3 Imatinib 0,1 1,3 1 1,3 4 U0126 1 1,2 10 1,5 5 ROCK inhibitor 1 0,8 10 0,7 6 DBZ 1 1,1 10 0,6 7 Ly 10 0,8 100 0,7 8 CHIR 0,1 1,1 1 1,1 9 Cyclopamine 1 0,9 10 1,3 10 PP2 Scr inhibitor 1 1,1 10 1,2 11 Bay 1 1,5 10 0,8 12 PD98059 1 0,9 10 1,4 13 SB203580 1 0,6 10 2,1 14 Roscovitine 1 1,8 10 0,4 15 JNK inhibitor II 1 1,4 10 0,8 16 Necrostatin 1 1 1,1 10 1,1 17 Pi103 0,1 0,6 1 0,7 18 FTAse 0,1 0,9 1 1,1 19 Tyrphostin 0,1 0,6 1 0,8

metabolic 1 Rapamycin 0,01 1,2 0,1 1,2 2 Metformin 1000 0,9 10000 0,7 3 AICAR 0,1 1,0 1 0,7 4 SCD-1 inhibitor 0,1 1,6 1 1,1 5 DGAT 10 1,4 100 1,1 6 Compound C 0,1 1,2 1 1,0 7 CsA 1ug/ml 1,2 10ug/ml 0,8 8 MTX 1 1,2 10 1,4

vesicle/trafficking 1 Tunicamycin 0.1ug/ml 1,1 1ug/ml 1,1 2 Chloroquine 10 0,8 100 0,7 3 3-MA 200 1,1 2000 0,9 8 4 Concanamycin 0,01 1,2 0,1 0,8 5 Leptomycin D 0,001 1,1 0,01 0,7

transcription/translation 1 17DMAG 0,01 1,5 0,1 2,5 2 CHX 1 ug/ml 0,4 10ug/ml 0,1 3 ActD 0.01 ug/ml 0,8 0.1 ug/ml 0,1 4 2-aminopurine 100 1,5 1000 1,4

epigenetics 1 5-Aza 0,1 0,9 1 1,0 2 Zebularine 10 1,2 100 0,8 3 Vorinostat 0,1 2,0 1 3,9 4 Panobinostat 0,001 1,0 0,01 3,7

proteases 1 LACTACYSTIN 1 1,0 10 0,7 2 CALPAIN 1 0,9 10 1,5 3 DC1 1 1,3 10 1,5 4 MP1 1 1,1 10 1,1 5 CA074 1 0,9 10 0,9 6 Pepstatin 1 ug/ml 1,7 10 ug/ml 1,2 7 AEBSF 100 1,3 1000 1,4

DNA repair 1 Olaparip 0,01 1,6 0,1 1,5 2 NU1025 1 1,2 10 1,8 3 NU7441 1 1,1 10 0,3

170

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 170170 008/05/168/05/16 22:0622:06 Chapter 8

Supplementary Table 2: Genes regulated in CSC signature, highlighted in blue are the genes with lowest expression in CSCs in orange are the genes with the highest expression in CSCs compared to differentiated cells. Log fold change (LogFC) is the change in relative expression between A) CSCs vs diff cells, B) CSCs vs CSCs + panobinostat and C) diff cells vs CSCs + panobinostat

)'.,)0$ , #+"0((()(.)0 0"*'(.**-,, '"''(-,-+,0 #'"**(/0( ((",0(*+0 '",/,/)'. #'"-,.-.0 -",(.**** '"-,+(/./ )'./(+$ - #+"*+*,+*+,- 0"(+./*0*'( '"''(())(+( #'"+/.,0 (("'.,/(- '"+(,)..- #'"(0))+* -"/.00+,0 '"/,0,/-- )),'(-$ ( #+"*((')-.( -"0*+(*.0'- '"'''.',+'. #'",/'.-- /".00)-/+ '",*0+*,. '")-(-,+) +"0'0+,(- '"-*)+()- )*((--$ (,, #+"'''''*,+ ."'(*0/+-,, ("(+#', #("(*00.* /"++*0000 '"''-'--* '"0-0)0(( ,"+0/-)/+ '"''+/.0+ )(.')/$ + #*"*',-0-.'- -"'*,+)+))0 '"'''(.)+(/ )"'*-++(/ /".'-+0*, '"''*(00, )".0*)+() ,"..0(0-, '"'''())+ )',/(,$ * #*"(00*0(,.) 0"'+...(.-( '"'''(.)+(/ #'"./*,0 ('"),,-.* '")'(.(), #'"++/)*+ ."))*0,/0 '"',*+(), )'0'*($ ( #)"/'(/,*+*+ ,"+.0.*.(,) '"'''(//+), (".()**-) .".*-/*) '"''),-,0 *"*,+).+0 ,".,,0+.0 )"((#', ))..')$ +( #)"-'/+')+*+ -",+*/,)0.0 '"'''**/.(( #'"(('/(/ .".0)-+,) '"/'//.-0 '").0/0'. ,"*.0,0.( '"*/(//+, )),-//$ $ ) #)"-'(/).0*. -".)-((,)() '"''/()'/*( #'"*.)0(/ ."/+',. '",))0,), ("',-,-(0 ,"0,*+/)) '")''./,/ )',*,)$ ( #)",)(.')-/0 ."'0((/*'(- '"''(((()., ("++0*/(, 0"'.-.),( '"'(),.(- )"/.(,*+( .")--'0/. '"'''--.( )()0,-$ (0 #)",(--'+(0+ -"(00'//)* '"'(0'(.0.. '"0+'(')- ."0).++(- '")'''.*) )",(,,*00 -"(0/,,-( '"',+,(.. )(0(.0$ ( #)"++'-*.,,( ,"/-+,+,--) '"''',(00/* '"-'-'0'0 ."*/.0'00 '"(.().++ ("(*).00- ,")('-)-. '"''*+*,- )*,/+,$ , #)"*,'+**))) ."-/0/))/-) '"'''(,,+)* '"0-(*(-* 0"*+,-0.- '"''.,0 ("*',-*(* ."(-.+)(0 '"''-+,/0 )(/0.+$ #)"*+,*-.0,- -"',/'-),*, '"'''.',+'. #'"'/*'(, ."(/0)*// '"/+-.,,- '".-+'('/ ,")-.*/*0 '"'*'-'.) )'+'+($ #)")0)('-'+- -".(*(*.+.* '"'''(,,+)* ("'*)),0/ /"*.,*)'+ '"''*.*') '"/)+'',) ,"0.0'/. '"'(,.0-, )*,)('$ $ #)"),/(.*+0+ ,",(0/(0(,0 '"''(),+'(/ '",(,().0 -"0'-+-00 '")'/*)-* '".*/().+ +".,0.0-( '"('00-*) ))-*'*$ , #)")',/-)+/ -".0')+0,- '"'''.',+'. #("))/0/- .")./-/./ '"''-'))/ #'"*+()*/ ,",(--00- '"+,0/+-+ )(()'.$ $ - #)"(/0/-,)(0 ."'.+'.0.-. '"''(),+'(/ #("'',*(/ ."---*,*, '"',).+,+ #'")+'./, ,"/,/.,+, '",0().'. ))-./0$ #)"(.,0,.+'. -"'0',)++*( '"''',(00/* '")+//()- ."*')0'0, '"*',+/'0 ("(,(.)+( ,",./+'./ '"'()0(0, )'.'(($ $ . #)"'-(*.(/)* ."'/--')-/. '"'((,0.*0+ #'",'+..0 ."/-+/00* '"+/'-,/* '"*.)'..0 -")+(0,,. '"+,(0'*. )'*.0.$ ( #)"'('-.',( /".0(,''.(0 '"'')('()// #'"+*)0+( 0",/'*-,, '")'/*)-* '"-.,.)'+ /"()+'),. '"(/-//. ))(')0$ $ , #("0/++-,')* ,"/*.('.',/ '"'''(.)+(/ #'",..-'0 -",+',*,( '"'+,,/,. '",*(',,0 ,"(('+'), '"'.-(/ )'-.-,$ ) #("0.+0-**-/ ."()/))(/'0 '"'')'+.+,+ '"'''(''/ /"((,.,*0 '"000/))+ '",+-',., -"+(*.-/0 '"),/0/, )(+')*$!$ ) #("0*-/-/'*0 0"0).(0-0), '"'''.',+'. '".-,,*,. ((")./*00 '"')*'-*. )"(*/.)/, ('"')/(). '"'''.-.( )',)./$ ( #("///(.+0,0 -",(()'0,,- '"'''(-/',* ("'()/'(+ ."0-(-0./ '"''*(00, ("(.,',.( -"(,+-,'- '"'''.-.( ))*(-/$ #("/./)*).() .".)()/+0** '"'''./0*( ("+''+,*- 0"*-'-)/( '"'')')++ )"..('/( /"(-..'0( '"'''+*+, ))0*.-$ ( #("/.,*/*)/ 0")),)..//( '"'''.',+'. #'"*----) 0"0.0-*/, '"*-/,-(( #'"'0+',, /")+',,0 '"/*-+++( )()*,+$ ( #("/.'//+,(, -"+-))0)/-+ '"'''(.)+(/ #'"/)))/0 -"0/-,0'/ '"''-0)(( #'"'/.0, ,"+/)/.,+ '".(+,++/ )*'*.)$ ) #("/'0''.*.( ,"(0,.)+-0( '"''',-)''+ #("((0'*+ ,",+'.((- '"''-'--* '"'0+-)// +"**/,*,+ '".*0+++/ )()/,'$ $ + #(".+-+-,+-- 0"-'**',0'. '"'''))*0'* #'"/(-,0/ ('"'-/)+ '"''-'--* '"()0.)0, /".0+0*.0 '",+0(++/ )'*+*+$ $ #(".)/-+(/*) +")/0.()/,0 '"'')'+.+,+ '"*-'+-*/ ,"**+)-,. '"+*+0/', '"'+0*',, *"+,''++. '"/+),-(, )),0(,$ *0 #(".)))'.(0* 0"0/)).+(/0 '"'''-)/.0/ #'"0('*.. ('"*//(/0 '"''+).', #'"()),*- 0"',00')+ '"-.+./0- )*/')($ $ #("-,/-.++(* 0"*,+*00(++ '"'''*('0*) #'"*0.). 0"0/,('(+ '"('.0../ '",'(()-( /"..,-), '"'***-/. ))/)0*$ . #("-+-.(''*- /"+((,0--)+ '"''')/,00( '",-('*). 0",(,+-/ '"'*+*,'/ (")0+',0( /")*,).(( '"'''.-.( )**0+0$ $ . #("-+(0,).(/ -"-'+*/-,), '"''(),+'(/ #("+-+,'0 -"-0*('/, '"'')'00. #'"(*0+)/ ,".(*-0- '",0-)(/- ))0-)*$ (,' #("-))-+.')) ,"(+.0).,-/ '"''(),+'(/ '",',')). -")((.-)+ '"(*./0.- '"-).'(-0 +"-,'((), '"'.+(--* ))''--$ ) #(",/,.,+-') +"/-.(0(..0 '"')*,+)0() #'"'-(/, ,"-)0(++( '"0)'0(-( '"+,'/)0 +")00.)0 '"(+''.(0 8 )),/(.$ ( #(",-**)0//( -"''()((/-+ '"''(''/-0. '"+'/(''. -"0/-0).) '"))0*,'/ ("'+0,,)* ,".++*)*( '"'()0(0, )'),'+$ )0 #(",-(*0','/ -"-'*,'+*(. '"''+'+*,,, '"*,.',)( .",-).),- '"+-)0.(/ '"---//,* -"(,-),(. '"'*0/0-* )(+*,.$ ((', #(",-((-())/ -"0//0+0*) '"''(+*//)+ #'".+/)+ ."*0,+( '"'(),.(- #'")'-)0 -"(',))*, '",0/''-( )')+,/$ )* #(",+**/-0() -"0'+*0-,// '"')**-)*, '"/++)+,) /"'0/)()- '"()+/00) (",,0)*-/ -"0()*)(, '"'(,/ )(('.($ $ (( #(",'-,**+(. ."/*''-..*. '"'''-)/.0/ '"'-(/(00 /"-(+)+++ '"..)),-. ("(,---0+ ."-,,(*,. '"''*)**, )'0/(,$ ( #("+0.),+)0- /"00'*')-0, '"'''-/0)-/ '"('+-).. 0".0()+*. '"-0/())( '".0/0*-. /"-+((+*0 '"''0--)- )(,().$ $ ( #("+0(*+.,0* ,"+,,--*+-. '"'''.',+'. )"*'+'+.* ."*,**-'0 '"'''),/ *"+/0+.*+ -"+,+.)-+ )"((#', )'*)*)$ $ ( #("+,+(-'()* ,",)(0'.,00 '"'''.',+'. '"+0)),(. -"+0,((*, '"'/**(0) ("+./.),- ,",*+(0'* '"''*(+// )'--(0$ + #("+*--((+'( -"*.0,+('., '"'',00.*() )"*.(/** /")/*.-** '"'))/+,, '"*0))-, ,"/,.*-.0 '")-(-+/ )''0/)$ $ - #("+*),)0.,- .")))*-')(/ '"''--/'+(0 '"-.*'-*/ /").,(,. '"(.*-)- ("*)0)/., ."(.'.*0( '"')')/+0 ))(0(($ ( #("+'(0,*,.( -")*,/0/-*) '"'''+(.)(. ("*0'-*+ ."-*)(0)+ '"''-'--* )"*)-*/,- -"-0/((+- '"''')0/( )),-).$ $ ( #("*00/)+00. /"+/../*/(+ '"'''.',+'. '")(.00.( 0")0--0+0 '"+(,)..- '",/--,,( /"'/((0/0 '"'*').,- ))(,,/$ $ ( #("*0''*-+*. 0"'+**00),( '"'''0*'..) '"+++.-. 0"0-'/'( '"'--,(0- '"**)0(( /",(+/*-, '"),/0/, )'+.*($ * #("*/)+-(()/ ."+.0(/,.,) '"''+/,,'+) #'",).++( ."0'--0,0 '"(+),)*0 '",*.-*-0 ."',-..*. '")'./*/( )('.).$ #("*..0,..,. ,"0,)'(.*/* '"''+')('*) #'"')))() -"-)0/0', '"0,.+('( #'"'(/+0, ,"),*.0'/ '"0*.+-+- )*(/00$ *() #("*-.-+**0) +"0)((.)/*, '"''-+0-** #'"++)/0+ ,"*/*,+.. '"*+',,), #'",,',,- *"0-)'.*+ '"')/.'*( )++*,*$ $ )() #("*-')(,..* -".+-**//(( '"''(*'*/** '")-+.+,- .",,//(0, '"*0+)/0- ("()(..,, -"-).((/. '"''++'/) )'0//*$ ) #("*,-*(/*0/ .")00,.(-.) '"'')'-'.0, #("(.+'), ."*0'.(/) '"'('-)0( #'"-*.'*, -"*')/0+/ '"'0**()) ))*)*,$ $ ) #("*+,(+-*,) (("0.0(,.0, '"'''--'// #("/.*-,- ((".(+0'* '"'''),/ #)"+'-.+/ ('"('*)(( '"''')-*( )'0/0'$ , #(")0+0(*'(+ 0")(***(,- '"''-+0-** '"')0(/.* 0"/.,*/(. '"0*+/00 '"+*./**( /"./+.0(- '")(-.,). )),*,,$ ( #(")/0+/.*,/ -",*00**'./ '"''+*(,(.( (".-)*,+ /"'-,/,*/ '"''-'))/ )"',,+,*. -"0))0(-) '"'''+)*+ )(''0,$ $ * #("),,(.''0, ."'-.)-..-- '"'))(',.() '"+/-0,-, ."0*/**(( '"*-,0.-0 '"+**-((- -"-,-+//, '",+/*./. )).(0,$ ,'* #("),*'.0+)- 0"/*0','.,- '"'''.',+'. '"'-,//-) ('"+0/,*+ '"..0,00- '".',,-'* 0",-,)0() '"''0.++0 )',0/*$ ( #(")*0*.,,-- 0"+((/*'-/* '"''..',)0* #'"('*).( 0"0.0//). '"..0,00- '"'-('-/( /"/))-.. '"0(,)')+ )'+*-+$ $ ( #(")*'('*(-/ ,"+/*,,+000 '"'''/+))0) )"*(*-,), ."),,+*)/ '"''(*),. )"-))0,+, -"(.00/'. '"''')0/( ))0+/($ ( #(")'+0()*', -"-/0-.(/.* '"'''.',+'. #'",')/.( ."'+'-0)- '"'-)/-./ #'"))0*0* ,"0.),(0) '")-,+,+ )'++(,$ - #("(0-*/0'*, ."'0/.)-(** '"'()())+'( ("-/+',. /",*/0+0) '"''*(00, )"(,./(/( .",.0++'- '"''(.*'/ )'-)*0$ $ ( #("(0+()**). /"+*+0.).(. '"''.)/)/,( ("'/+,-/ 0",.+*(/+ '"''.),(. ("+'-''.( /",+'0(+- '"''-(-++ )(')*0$ , #("(0(*0.(') -"(,)/+* '"()'/,(0/. #'".'-(++ -"*0,+-0/ '")+'),., '"''.+,,/ ,",-'/.)* '"00+'0/- ))-''($ , #("(.,(-.*)- ."0)'*0/+,+ '"()(,0.0* '".'00/.0 /"/-)0.-( '"('*('+* (".'.'+() /"(/-**,+ '"'-,,.+* )()0+)$ $ #("(,*)-()+. (("'0,,-//, '"'')-'///. #'"+**/,) (("+,,).+ '"(*++,,. #'")')()) ('"+(./.. '"+,0/+-+ )(*(.)$ 0 #("'/)-0((/, -"'+)/(-'.0 '"'',00.*() '"+')+-,0 -"./,*0+- '")'/*)-* (")0++-+, -"(+/.')/ '"'')(.'* (,--,,.$ )#( #("'-'-00,(/ -"'/+*-'*') '"'')-'///. (")0'/(+. .")-'((.+ '"''+).', )"'*+.), -",.(*.* '"'''.-.( ))/'++$ ) #("',**)00) /"*.*+',-+( '"''(''(/- #'"-.,,+/ /",-))0-, '"'(/'0.+ '"'.'(++ ."//(/(). '"-.,*(() ))../)$ . #("'*/00./0( -"//,-.(-// '"'')**'*// '"-(/0),* .".(+-*** '"')*'-*. '"+('+/-- -",.(+(- '"(/0..+- )*,*'/$ )' #("'*/)(/*0+ -"(.)/0(', '"'',/').'. '").*/'/, -"/)/0'+, '"*-,0.-0 '"*/,*+,0 ,"/+-+,+/ '")'/.),- 171

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 171171 008/05/168/05/16 22:0622:06 Chapter 8

))/'*/$ ) #("')+(+/0(+ ,"-/,'./..( '"''+)'/(0. '",(-).0/ -"+,,)0*( '")..00/ '",,.,/(+ ,"+,(.0, '")''./,/ )*,.+-$ $ )( #("''(,)/0(/ ,"'-+*)(0,- '"''*+.00(* #'"(0/+ ,"+-,//-- '"+./+/). '"/.*/,00 ,"'''+/.+ '"'',,*') )'*'*.$ $ ( #'"00(',-(.( -"0'++'0/0* '"'')-0*00- (")*0-',. /"'(0.+'/ '"'')'+-) '",0))//. -".',')-( '"')/.'*( )')--'$ ) #'"0/.,,'.(* /"'-.*/0-*/ '"''(+-0*-, '"'.'0/'/ /",0--,,+ '"/(*)'/- '"-***+,) ."/0')/-0 '"'(-'++* ))-/)'$ *-) #'"0,+,.'/., ,",*'*0(.'. '"''/0-'/). #'"-)**,) ,"-0-''( '"'.'()/- #'"(/')0- +"0-)0,/+ '",0-)(/- )')0.+$ ( #'"0*''(++0, .")*.***,(- '"'('+,)-)/ '"/).0(.* /"((-)00+ '"'(-.)*/ ("**-+)), ."++',*., '"''*)**, )(/,.+$ $ ( #'"0((.-'(*0 .")./+-(*0) '"'')-+(*.) '"*.''./- ."0(0*/'. '"(*('*/( '"0'0',(, .")..('.( '"''*)**, )(+('+$ (-( #'"0'-,0*0+. -"/0-.,-(+* '"'')))-//- #'"''(,-. ."*+0)-00 '"00./.0* '"-',-0'0 -".+-*'+- '"'(0,+// )*0-+.$ (* #'"0')---/*- ,"*))','*'* '"'(+0()-*- #'"(.//)( ,"-/*0.*( '"-*0()/( '"((.,'(* +"0)0+-., '",*0*.-( (,,,0*,$ $ #'"//*/*,-.. +"++)0),',. '"''/-0''') #'"++0+/0 +"--''0/, '"((+00)) #'"')-+ *"0/./'.* '"0+.,0'( ))/+-($ ** #'"/..--,*'- ,"-0+-0.*() '"''--/'+(0 '".,)///0 -",'00.++ '"',,*/*/ '".*0,*0/ ,"-),-*+, '"'(,.0-, ))00).$ ( #'"/-/-00*/. ,",-.0//-(. '"'(.+')-0( '"/.,'**) -"+*0/,+0 '"'*+*,'/ ("))'.//* ,".++'**( '"''-+,/0 )'(-+0$ ) - #'"/,--+,(,, -"+))-00/+/ '"''-+0-** (")+/)'(- ."+.,()*) '"'')'00. (".)'+0(, -"/,+-)* '"''')0/( ))0-+0$ * #'"/*)/-,/() 0"'-)(*0), '"''0'/,(-( '"'/'./). 0",(/0-*, '".,/(0). '"(//---) /".+''*0, '"+,0/+-+ )(*//'$ , #'"/*',+.**( ('"/*0),/*) '"''+///.0( #'"0+0--( ('"..0.'( '"'(*0-.( #'",)*-/( ('"(-)(++ '"')--,-+ )'*',)$ ) #'"/(+*+(((0 -",,-++'+,) '"'()/',,-. #'").)0,- -"/).(** '"*(,-,-* '")/'(/.- -")/0*-*. '"*0/('** )'0*(/$!$ ( #'"/',+),,'( +"+,0(--+- '"'',/').'. '",0+'(,. ,"(,///. '"',+/()) '".-.-)+* +"++')-,0 '"'()*,0( ))/)0($ $ #'".-/*.,+,- 0".*.+'(0/* '"'(.),+/+- #'"(0)-)) ('"'),).0 '"-('/'.* #'"('+++/ 0"*''00'( '"-.+./0- ))''+/$ #'".-.)0.-'- /"(,,/('0), '"''/-0''') '"+,-/+,. /".-.//)- '"(().0.( '"./(()-+ /"(-).),* '"''.-))/ )((-.,$ $ #'".,,+()), -"'++*,,(.- '"''/0-'/). '".),-0./ -"./+0(') '"'))0'', '".*---./ -"'*+0/* '"''0/0.0 )).-+)$ ) ( #'".+0-('(*. -"'-..0+**/ '"'(,*+-+0. '",(/-*/- -".'(0(/. '"'//')/- '"-,**/)* -"'(0-/'+ '"'(,.0-, )*))*($ ) #'".''/0-)/) -"+/0,+'.- '"'0/,,.--( ("((-'/,* ."*0/'*(, '"''+**,* (".+/-(-0 ."'(*+'(( '"''.-))/ )),.)/$ ) #'"-0,/)-/(+ -"/*)*./,0. '"'-0*,(/'. #'".(,',) -"/)).--( '"(*-00+0 #'",')..* -")**'.// '")(-.,). ))/)()$ ) #'"-0,+..0) ,"/.,0..)(* '"'*'-*.(// #'").0*0) -"'/+')') '"+'''. '"'*(**- ,",+*0'-* '"/0'+'* )(+)-($ $ - #'"-/+00'*(- ,"*'*/.(*(* '"'*(0/(+(. #'"**,,/) ,"+./,.,, '"),-*)// #'").*)', +"/)+..*. '"**../*( )',/*0$ $ ( #'"--..,(-,/ -"'*+)-0.'0 '"'(())*(*/ #("((/0', ,"/'/-0)0 '"''*/*/, #'",',-.- ,"++.,,- '"')/-*)/ ))0-*/$ * #'"-,),''+., -"(/'.,*(-. '"'()())+'( '"+*-/,.0 -".),+*)* '"(*+'.*0 '"*(+*'/) -"'((-,. '"('++0+, )'00*'$ $ ) #'"-+.,0'--- +"+-0,(+',* '"(--/+-,/. #'")-*/0, +"--(*-) '"-*./(/0 '"'*'0-*) +"(-()''* '"/0'+'* )(*-*/$ ( #'"-(,),(--* +"(,,-+('), '"'(0,/0)/. '"'.'++'. +"+0/+/.) '"..0,00- #'"'*'.+0 *"/*)-+', '"/0'+'* (,-0-)'$ $ (''()/'.0 #'"-'+*0((*/ +"0-.),)/( '"')..-0'- #(")--*// +"-*-),+) '"'')'00. #'".,-(,/ +")/-0./* '"''0--)- )('))'$ ) #'",,--0/,/+ ,"0-,*++'./ '"'-/-//'- (")+),+'/ -"/-+0-*/ '"'((',++ (")',,'-- -")/0.+/( '"''0--)- (,,*0.($$ #'",,+'.*./+ ,"+,)'0,)*) '"'++*+)/)( '")0')/'. ,"/.+).), '"*00'*/) '"+'*)+-0 ,"*.--/(/ '"((/(*() )')'/-$ ( #'",*,,+).0. ,",,,'/,.0- '"'(.')'*', #'")+',*/ ,".'),//+ '")/),-., '"*0,*,*( ,"+/+00'0 '"',+(,,* )+)(*/$ ( #'",'0//**,. ,"((+',0)0/ '"((-/+-)/+ '"0+0/.,* ,"/+*0*/- '"'(,-'), (")'-0*00 ,"+-),/.- '"''/*--. ))/-+0$ #'",'//'*0.( -"--.'-,/*0 '"'/-,')(,) #'"(*.--. -"/,)-*+* '".')(/.0 '"'0++,- -"+,0/0(/ '"-,'0-(( ))'+-)$ * #'"+0+++.))- +"+*,0.(0,+ '"'-.,/.)-+ '"*((,') +"/*/0+-- '"*.-,+-* '"*)0/,/0 +"*,*-../ '"(',)+'. )*,*-'$ * #'"+0')(+0*( ,"',.,((0*, '"(0((0'00/ ("')(/.)* ,"/(*,,,- '"')*'-*. ("(.'(,*) ,"*0.+/(( '"'(+/.,/ ))+,(/$ $ ,,0 #'"+,)''*'-+ -"+,/+*)-./ '"*'+00,/./ '",,,*/(. -"0-)(),( '"(-,00/( '".')).') -",/*,--* '")''./,/ (,,*((,$ ( #'"+(-('**)+ -"(-.0-*(,0 '"'/.,.0-.* #'"),*)'. -")+0+((* '"*0'()+0 #'"(*).0. ,"/0*,(* '",-'(') )'-(+'$ ) #'"+(,.'0,/, ,",)+00**)/ '"'+,-+,)). '"0'))((* -"(/*0,*/ '"''-0)(( '"-'0(,. ,"-)(.(. '"'(0)(0* )'0.0+$ * #'"+(+.),/0+ ,"0.,)/(./+ '"'.0)(00). '")'-*0., -")/,/+*, '"+()0')+ '")0(+-+, ,"0(*-,(( '"(/0..+- )*'+/0$ , #'"+'+,+)++) ,"/+.'++-*- '"'.)(',.,, #'"(+'*0 ,"0.0()( '",/,/)'. #'"(+-)'/ ,",.(--0, '"+/,.)(* )')+(($ ). #'"*-(+)-'') ,"0)+0.+,)- '",+/,)*+( '",0.-)(- -"+'++0/* '"+./+/). ("'(0)'.+ -"),*/-,) '"(/+0)0, )*(')+$ ,#( #'"**+00))** ,"(-(/-/+,. '"('()+,)*. '"(*,././ ,"*0.),/, '"-*0++0+ #'"'.*/'+ +"0,.+.'- '"-.0./)* ))-(+,$ $ ( #'"*),+).(0+ -"/)+/-.'*( '"(,)-'*+)0 (",(//)+, .".+-00)0 '"''(*(0) (".+.0**) .",*-()'( '"'''.-.( 8 )')(**$ ( #'"*)')+0')0 -"-+//,*-(- '"+.0*++/)0 (".-'/(// ."-/0*/., '"'')').) ("*-0.,,/ ."(.*-'. '"').(+. ))-,,)$ , #'")0(+/*(-0 -",--/++),* '"()0*)00*- '",..*( ."''()+'0 '"'**0-,0 '",*.+,.. -"-/0/*(, '"'(-((0/ )'-))+$ ( #'")/,)+/,-+ ,")+,00/-+/ '"*-,''-,)) '"'/0).*, ,"+**),0. '"/*+,(,, #'"()++(+ ,"'+((-.( '",*0*.-( ))(+-*$ )+ #'").-'.'0)) ."+)+-,)*( '")0',,'((0 #'").,--, ."+)+/,,* '"*-,0.-0 #'"*,+.,+ ."('0)*0- '"'.-(/ )'*+)*$ ( #'").+).0(,+ ,"-','/.*)- '")('+*+)0- '",(-'(0* -"''')*-- '"()/+.-+ '"',)(*.) ,"+0+'(-* '"/'*'*+( )'**',$ (*( #'").(0++-(, ,"'(*0.(/,, '")/'-)/+)- #'"'.+./- ,"((),,() '"..0,00- #'")*--,+ +".,0-.)+ '"*//('-) )').'0$ #'")-.,+-/(* ,"-++*(),(. '"(.'0-(-)( #'"(+,(*- ,".',,(.0 '",''.-(0 '"'(+++(0 ,",(..-'( '"0,.)+-0 )'+.0.$ $ ( #'")+)//0** +"*0'./)'.( '"(0,((,,+* '"0/-)('0 ,"'',**)) '"'-)/-./ '")*0*'// +"*//00(/ '"(0.*,,) )(*,(0$ $ ) #'")(/+.,.)- -"*),-)),. '"*((')0**( '",.,-'+0 -".))--)0 '"'-+)0(+ '"-*++.0. -",**-)+- '"'*'-'.) )',*+)$ $ () #'"(++,*,.-. *"***)-'*)0 '"+./)(,)/, '"(++.--/ *"+..0((- '",..'*0. '")'-/+-0 *"*-++(,0 '"),/0/, )',0('$ $ #'"()+.+(/', *"/0)(+-),0 '"+..(0'((* #'"('()(* *"0'*0('. '".')(/.0 #'"'++-.( *"/'.+*00 '"/),*+', )')0-,$ $ - #'"())/')()( *",+,'0),-) '"-).0/)*+) ("(-0+,0, +"(0())*+ '"''-'--* '")-(0.0) *"-(+-/(( '"(/+/0(. )''/,0$!$ #'"'-.-*,.-. /"(*(-,*)0. '".+//*(.0, '"++./.++ /"*/0+'/+ '"'/**(0) '",)'+).+ /"*,/'+0( '"'/,)*/. )*0((/$ ) #'"'++(*0*00 *"*-0./*++. '"/(+0*).*0 '"(-'/ *"+.)),*( '"+/'-,/* '"'(0.-), *"*,.,0, '"0(,)')+ (,,)-0/$ * #'"'**0)+.-0 ,"0/.)0..(+ '"/.+.,(/+( '"(0-'.(/ -"('))0- '"+')+/+ '"(/(0',/ -"'-()//) '"*+)+'.* (,,.)/,$ #'"'(+*-0(,, -"--+0.).0( '"0.()'+,*+ ("+).-('* ."*/,0-), '"'0(.((+ )"'/.()(/ .".'(*+0( '"')*,-)- )'.+()$!$ '"'(.)/(-0) ,"('+,,(-)/ '"0).'*,,'- #'"'/0(- ,"',(**'. '".')(/.0 #'"'0'/., ,"'-..,+/ '"-(+)*)- )*'.'-$ $ )) '"'(.0,,(,/ ."'0-*,0((- '"0).-/'+.0 '",//'.,0 ."*/(+(0, '"',,*/*/ '"+)0-,'0 ."*)'(-)) '"'*-+,- )),0-/$ ) '"'))'.''00 +"*('.'+)* '"0'-'/(0-. #'"'.*0)- +")-).'- '".*/,('- #'"),(-// +"(0,/0,, '"(0.*,,) )'---+$ '"'*(**+)./ )".,'-+-''( '"//*').(** (",*',)-) *",'')+) '"''-'))/ )"(*++/,/ *"/**,,- '"''')0/( )*..)($ $ + '"'+-0+),0, *"/')*.,0)+ '"/(+0*).*0 '"'*+,)/, *".0-(-/0 '"0'/.0./ #'"(--('* *".+).0,. '"+0-)/0, )'.)(+$ + '"'.(..)(,* ()".)/,,''+ '"/+,0*/()+ #'")('0)* ()",/.)'* '"+**./+( #'"()*(-0 ()".')/,) '"/)/-.-) )+'(-($ $ )' '"'0.(''*0( *")(..),.(* '",0,,'+',) '"./,00.+ *",-)(.+) '"'(+./. '"*-/-/.0 *"+,'-(0/ '"'0('(*- )'*0/-$ )'*0/-$ '"(,+,+*)+* ,"0,'')'(** '"+)0/0(,/) '"(+..)'( ,"0+--'/, '",)/(0)- '"'./,)() -"'--,,)* '"-.+./0- )*)*)($ (. '"(/**-,*/. ,"()'0..*0( '"+++)/'./, '"'0'..,) ,"'.+-/)* '"..'0*(+ #'")./-(* ,"'.**,*- '")+((--. )'**.*$ ) '"),/+(+(* +"+,--00**0 '")*'+,)++- #'"'*0/+ +"*'.,.)* '"/.,.),0 #'"(.+)-0 +"+0/..(0 '"+,(0'*. )'*--,$ ( '")-)..-** ,"0*,-)-/++ '"*--''(()) '"..+')'/ -"(0()+0( '"'),-*-) '",,*)-.. -"*+*-+/0 '"'0('(*- )'-(+0$ ) '").+-/-(*- -"'0/*'0./ '"(*(+*)(.) #'"))-+. ,"/+..*(, '"*)+))+, #'")(0-,) -"(),/)-/ '")''./,/ )'/),'$ $ ( '")/000'),- ."'.-0.+)* '"(/0()/,+* #'"(+-**- -"/,//((* '"-*0()/( '"),.*'+) ."*,'-)(, '"*/(//+, )',,*/$ ) '"*-0+)*+-( ,",.-)')(*- '"))-*+('() ("-+0(0+/ -")(-'/./ '"'')').) ("0-+.0.) -".+**(), '"''')0/( )'++)'$ ( '"*0,',(,), -"/*0*(-*'/ '")/+,/*++- #'"(/0)+ -",+.(.'* '".')(/.0 #'",*)-( -"..',*. '")/0/-0- )(*,/0$ $ * ( '"+)),0)((+ -"0,)0/*(., '"',,0'+/,0 #'"+(,,*. -",**0(/. '"'0/'/-/ #'",*+-) -"/0-0-0+ '"'*-*'*- )*/'-*$ (,+ '"+-0(*'+-) -"(,'(+*(0. '")-'*.+'.* ("/-',((0 -"/+,/**0 '"''),-,0 (",,,*+ ."(-)*./+ '"'++'('. )*0).)$ )/ '"+.(,0-/, +"/.+--*.-, '"'0/,,.--( '"0)*(',* ,"(''+(/ '"''.,0 (",.+0)-) ,"/0.0),* '"''((-00 ))**/,$ )( '"+0.0)')00 -"+*0-)/'/( '")++/,).) #'"'*,')0 -"(.*(,*+ '"0)'0(-( '"''.0+)0 -"-0),,0. '"00+'0/- )()(0'$ ) '",+.-+)(+0 0").+/'+-(/ '"(.'0-(-)( ("'.)**,+ 0",*.(,() '"')*'-*. '"//0*,'* 0"00**''0 '"'*/*)*, )''/./$ ( '",,'/-*/(* /",/'-00')* '"'+/-)***0 (")**((/0 /"0)(/)-- '"'')'00. ("'0.)*') 0"+'+.+- '"''*/(++ 172

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 172172 008/05/168/05/16 22:0622:06 Chapter 8

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

173

TThesis_CD_Zimberlin.indbhesis_CD_Zimberlin.indb 173173 008/05/168/05/16 22:0622:06