THE REGULATION OF PANCREATIC CANCER PROGRESSION BY CD147 AND
CYCLOPHILIN A AND THE STUDY OF INTERLEUKIN-8 DYNAMICS AND INTERACTION
WITH CXCR1 RECEPTOR PEPTIDE
by
AGNIESZKA ANASTAZJA KENDRICK
B.S., University of Wroclaw, 2005
M.S., University or Colorado Denver, 2010
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Structural Biology and Biochemistry Program
2016
This thesis for the Doctor of Philosophy degree by
Agnieszka A Kendrick
has been approved for the
Structural Biology and Biochemistry Program
by
Jeffrey S. Kieft, Chair
Chad G. Pearson
David N. Jones
Sean P. Colgan
Richard E. Davis
Elan Z. Eisenmesser, Advisor
Date: 05/20/2016
ii
Kendrick, Agnieszka Anastazja (Ph.D., Structural Biology and Biochemistry)
The Regulation of Pancreatic Cancer Progression by CD147 and Cyclophilin A and the
Study of Interleukin-8 Dynamics and Interaction with CXCR1 Receptor Peptide
Thesis directed by Associate Professor Elan Z. Eisenmesser
ABSTRACT
Extracellular Matrix Metalloproteinase Inducer (CD147 or EMMPRIN), cyclophilin A
(CypA) and interleukin-8 (CXCL8) are proteins implicated in the regulation of several different cancers and inflammatory disorders. In each case, the specifics of their activity and molecular interactions are not fully characterized and were thus a focus of these studies.
CD147 is a highly glycosylated type I transmembrane protein upregulated in pancreatic cancer where therapies are being developed to specifically target CD147 and its subsequent detrimental effects. Such therapies have not been fully effective, presumably due to the lack of information on the molecular details of CD147 activity and interactions.
The present study identified CD147 in having a unique role as a chaperone, for several transmembrane proteins important for signaling events related to cancer metastasis and metabolic maintenance. In our model CD147 depletion led to decline in cell proliferation, increase in epithelial phenotype and striking reprogramming of glucose and glutamine metabolism. We further identified several transmembrane proteins to be in a complex with
CD147 and the specific mechanism of CD147 activity was characterized in regard to its glycosylation status. Together, our studies illustrated the functional influence of CD147 expression on pancreatic cancer progression and unravel a novel mechanism of CD147 regulation of cancer through its chaperone activity.
CypA, a peptidyl-prolyl cis-trans isomerase and CD147 interacting partner, is involved in protein folding, trafficking, assembly, signal transduction, and cell cycle regulation. Although initially discovered as a cytoplasmic protein, CypA has recently been shown to be a key player in the regulation of signaling pathways due to its highly abundant
iii
extracellular presence. We sought to identify the specific mechanism of extracellular CypA activity in regard to pancreatic cancer and discovered CypA to be internalized by pancreatic cancer cells via clathrin mediated endocytosis. Current studies implicate endocytosis as one of the emerging pathways important in the control of cellular homeostasis and proliferation.
Our analysis reveled that this cellular uptake process is reliant on CD147 expression and not dependent on CD147 glycosylation status. Here, we describe novel evidence for CypA internalization by cancer cells and thus, provide a possible mechanism of CypA signaling in regulating disease progression.
CXCL8 is a pro-inflammatory chemokine important for the regulation of inflammatory and immune responses via its interaction with G-protein coupled receptors, including CXC receptor 1 (CXCR1). CXCL8 exists as both a monomer and as a dimer at physiological concentrations and several biological studies have indicated that both the CXCL8 monomer and dimer are active. However, biophysical studies have reported conflicting results regarding the binding of CXCL8 to CXCR1. To clarify this problem, we expressed and purified a peptide (hCXCR1pep) corresponding to the N-terminal region of human CXCR1
(hCXCR1) and utilized nuclear magnetic resonance (NMR) spectroscopy to interrogate the binding of wild-type CXCL8 and a previously reported mutant (CXCL8M) that stabilizes the monomeric form. Our data reveal that CXCL8M engages hCXCR1pep with a slightly higher affinity than CXCL8, and that CXCL8 does not dissociate upon binding hCXCR1pep. These investigations also indicate that CXCL8 exhibits inherent flexibility within its receptor-binding site on multiple timescales, which may help explain the versatility in this interleukin for engaging its target receptors.
The form and content of this abstract are approved. I recommend its publication.
Approved: Elan Z. Eisenmesser
iv
I dedicate this work to my brother, Paweł Baran
v
ACKNOWLEDGEMENTS
I would like to thank my parents, Elżbieta and Stanisław Baran, for their never ending love and support. I would also like to thank my brothers, Piotr and Michał for always being there for me. I would like to thank Michele for all her support. I would like to thank James, my husband and best friend, for the encouragement and support. I would not have come this far without his love and belief in me. I would also like to thank Ian, who puts a smile on my face every day and gives me the strength to never give up.
I would like to thank my mentor, Elan Eisenmesser, for his guidance through the years and for sharing his excitement for science with me. He has taught me how to be a better scientist and to always keep going. I would like to thank my previous mentors,
Douglas Dyckes, Karen Jonscher, and Sean Colgan, for allowing me to explore the field of biomedical research and teaching me how to critically think about science. I would like to thank, Mike Holliday, for his friendship and for being my support system over the years. I would also like to thank other past and present members of the Eisenmesser lab, specifically, Thomas Chi and Johnathon Schafer for their help with science, but also for making our working environment enjoyable and fun. I would like to thank the members of the
Hansen lab for their help with research and creating a fun and friendly work atmosphere.
I would like to thank my committee for their insightful comments and suggestions. I would like to thank the students, faculty, and administrators of the Biochemistry and
Molecular Genetics Department, and the Program in Structural Biology and Biochemistry, for allowing me to be a part of such a thriving research environment. Lastly, I would like to thank Sue Brozowski for her support over the years and Elizabeth Wethington for her friendship and ongoing care.
vi
TABLE OF CONTENTS CHAPTER
I. Interleukin-8 dynamics and its interaction with human CXC receptor 1 peptide ...... 1
Introduction and review of the literature ...... 1
Interleukin-8 ...... 1
Biomolecular NMR spectroscopy ...... 4
Study rationale ...... 8
Results ...... 8
CXCL8 monomer and dimer binds hCXCR1pep with slightly different affinities ...... 8
CXCL8 dimers do not dissociate upon hCXCR1pep engagement ...... 12
CXCL8 exhibits inherent flexibility within the binding region ...... 14
CXCL8M dimerizes at high concentrations ...... 16
Chemical shift-based calculations identify fast time scale dynamics ...... 16
Discussion ...... 18
Future directions ...... 21
II. CD147 regulates cellular metaboliSm via interaction with small molecule transporters in pancreatic cancer ...... 38
Introduction and review of the literature ...... 38
Pancreatic cancer ...... 38
Cancer and cellular metabolism ...... 41
CD147 ...... 42
CD147 interacting partners ...... 45
Study rationale ...... 48
Results ...... 48
The extracellular region of CD147 exhibits little to no activity ...... 48
CD147 downregulation in pancreatic cancer cells inhibits cancerous phenotype ...... 50
vii
SILAC analysis reveals differential expression of several metabolic and adhesion proteins in CD147 depleted cells ...... 52
MCT1 and MCT4 expression contributes to CD147 activity but does not fully explain CD147 mediated PDAC phenotypes ...... 54
MS based cross-linking experiments identify adhesion and transport proteins as possible CD147 interacting partners ...... 56
The conformation of possible CD147 interacting partners ...... 59
CD147 acts as a chaperone for several transmembrane proteins ...... 61
CD147 cellular interactions lead to the reprogramming of glucose and glutamine metabolism ...... 64
CD147 mediated growth inhibition is linked to metabolic reprogramming ...... 67
Discussion ...... 69
Future directions ...... 76
III. A new mechanism of Cyclphilin a internalization by cancer cells ...... 111
Introduction and review of the literature ...... 111
Cyclophilin A ...... 111
Endocytosis ...... 113
Study rationale ...... 115
Results ...... 115
Cyclophilin A is internalized by pancreatic cancer cells and the internalization process is cell type specific ...... 115
Clathrin mediated endocytosis is the key entry pathway for exCypA ...... 117
ExCypA internalization process is CD147 dependent ...... 118
The functional consequences of exCypA internalization are inconclusive ...... 119
Discussion ...... 120
Future directions ...... 121
IV.. Materials and Methods ...... 127
Reagents and plasmids ...... 127
Protein expression and purification ...... 127
viii
CXCL8 constructs ...... 127
hCXCR1 peptide ...... 128
CD147-ECD purification ...... 129
CD98hc purification ...... 129
GFP-nanobody purification and conjugation to beads ...... 130
CypA purification ...... 130
CypA-GFP and GFP purifications ...... 130
Mammalian CD147-ECD expression and purification ...... 131
NMR spectroscopy ...... 132
Assignments and standard NMR experiments ...... 132
Relaxation experiments ...... 133
Molecular dynamics simulations ...... 133
Mammalian cell culture ...... 134
Cell lines and culture conditions ...... 134
SILAC cell culture conditions ...... 135
Stable cell lines generations ...... 135
Transfections ...... 135
Immunofluorescence ...... 136
Intracellular organelle staining ...... 136
Proximity ligation assay (PLA) analyses ...... 136
Imaging and image analysis ...... 137
Annexin-V staining ...... 137
Cell cycle arrest analysis ...... 137
Cell growth and cell size analysis ...... 138
Cell migration ...... 138
Enzyme-linked immunosorbent assay (ELISA) ...... 138
ix
Cytokine profiling ...... 139
Western blot ...... 139
Immunoprecipitations (IPs) ...... 139
Proteasomal inhibitor treatments ...... 139
MCT1 inhibitor treatments ...... 140
CypA-GFP and GFP treatments ...... 140
FACS analysis of CypA-GFP or GFP treated cells ...... 140
Cold temperature internalization experiments ...... 140
Potassium depletion ...... 141
Receptor internalization ...... 141
Global metabolomics cell culture conditions ...... 141
Metabolic tracing experiments ...... 142
Mass spectrometry ...... 142
Proteomics analysis ...... 142
Metabolomics analysis ...... 144
Statistical analysis ...... 145
Binding isotherms calculations ...... 145
Metabolomics analysis ...... 146
All other data statistical analysis ...... 146
REFERENCES ...... 148
x
LIST OF TABLES TABLE
1. Dissociation constants obtained from chemical shift perturbation for the different CXCL8 constructs ...... 23
2. The proteins, their molecular weight and correlation time used for generation of graph in Figure 6...... 23
3. The extracted correlation times and corresponding molecular weights (MW) for the free and bound constructs from relaxation dispersion experiments acquired in the presence of phosphate buffer...... 24
4. Proteins identified in cross-linking experiments in PANC1 cells ...... 77
5. Inter-protein cross-linked peptides identification using Protein Prospector from cross-linked samples ...... 80
6. Antibodies used in the study ...... 147
xi
LIST OF FIGURES FIGURE
1. NMR chemical exchange regimes ...... 25
2. Sequence of the CXCR1 N-terminal region, and sequence and structure of interleukin-8 (CXCL8) ...... 26
3. 1H,15N HSQC NMR spectra of uniformly 15N labeled CXCL8 constructs utilized in the study ...... 27
4. CXCL8 engages human CXCR1 peptide weakly with increased affinity in the monomeric CXCL8 form ...... 28
5. The conformation of specificity in CXCL8 binding to hCXCR1pep ...... 29
6. Changes in chemical shifts upon CXCL8 binding to hCXCR1pep ...... 30
7. CXCL8 engagement of hCXCR1pep leads to dynamic changes ...... 31
8. CXCL8 engagement of hCXCR1pep does not lead to dimer dissociation ...... 32
9. CXCL8 is inherently flexible within the receptor binding site ...... 33
10. CXCL8M weakly dimerizes ...... 34
11. Chemical shift-based calculations correspond to fast time scale dynamics and reveal that the C-terminal helix of the CXCL8 monomer partially unfolds ...... 36
12. Graphical depiction of epithelial to mesenchymal transition (EMT) ...... 81
13. Metabolic reprograming in cancer cells ...... 82
14. Glutamine metabolism in cancer cells ...... 83
15. Schematic representation of different CD147 isoforms and function ...... 84
16. The activity of CD147 ectodomain ...... 85
17. Transfection and purification scheme for obtaining pure glycosylated CD147 ectodomain ...... 85
18. Purification of glycosylated CD147 Coomassie stained ...... 86
19. The activity of glycosylated CD147 ectodomain measured by MMP9 ELISA ...... 86
20. The Expression of CD147 and EMT markers in pancreatic cells ...... 87
21. Knockdown of CD147 in different PDAC cell lines ...... 87
22. CD147 knockdown is associated with reversal of mesenchymal phenotype ...... 88
xii
23. Migratory capabilities of CD147 depleted cells ...... 88
24. SILAC MS analysis reveals several differentially regulated proteins in CD147 depleted cells ...... 90
25. CD147 depletion leads to MCT1 and MCT4 downregulation ...... 91
26. MCT4 depletion in PDAC cell lines leads to CD147 downregulation but does not influence EMT ...... 92
27. Intracellular lactate accumulation after CD147 and MCT4 depletion ...... 93
28. Global metabolomic analysis of CD147 and MCT4 knockdown PANC1 cells ...... 95
29. Schematic representation of cross-linking/MS approach used for the identification of possible CD147 interacting partners ...... 96
30. Cross-linking/MS identifies several distinct families of possible CD147 interacting partners ...... 97
31. The conformation of the interaction between CD147-ECD and CD98hc-ECD is inconclusive ...... 98
32. Conformation of CD147 interactions via Western blotting ...... 99
33. PLA confirms interactions between CD147 and proteins identified in cross- linking/MS experiment ...... 100
34. Proteasomal inhibitor treatments protect CD147 interacting partners from degradation ...... 102
35. Restoration of CD147 expression rescues the interacting proteins expression and activity ...... 104
36. CD147 depletion prompts PANC1 cells to reprogram glucose and glutamine metabolism ...... 106
37. The loss of CD147 expression leads to several phenotypic changes associated with cell growth inhibition ...... 107
38. PANC1 shCD147 cells utilize glycolytic intermediates to generate cell building blocks ...... 109
39. The summary of the consequences of CD147 depletion on mesenchymal PDAC cells ...... 110
40. Prolyl isomerization reaction catalyzed by PPIase enzymes ...... 122
41. CypA is internalized by pancreatic cancer cell lines ...... 123
42. CypA is internalized via clathrin mediated endocytosis ...... 124
xiii
43. Internalization process depends on CD147 expression but does not rely on CD147 ectododomain ...... 125
44. The functional consequences of exCypA internalization are inconclusive ...... 126
xiv
CHAPTER I
INTERLEUKIN-8 DYNAMICS AND ITS INTERACTION WITH HUMAN CXC RECEPTOR 1
PEPTIDE1
Introduction and review of the literature
Interleukin-8
Interleukin-8 (IL-8, CXCL8) is a small (8.6 kDa) member of the pro-inflammatory
CXC cytokine family produced mainly by macrophages and in smaller amount by epithelial and endothelial cells2. All CXC chemokines are relatively small proteins (8-12 kDa) that contain a conserved CXC residue motif (C-cysteine, X-any other residue) proximal to the N- terminal region of the protein. Solution and solid-state structures of several human CXC chemokines have been solved, revealing a common tertiary fold for all members of the family consisting of an unstructured N-terminal loop, antiparellel β-strands and a C-terminal
α-helix3-6. CXC chemokines typically dimerize, and residues within the first β-strand and α- helix are important for the stabilization of the dimeric form. In CXCL8, dimerization occurs via an interaction between side chains in its first β-strand and is further stabilized by its C- terminal α-helix and two disulphide bridges (Cys7-Cy34 and Cy9-Cys50). Several different mutations or truncations that disrupt the structural order in these regions can lead to formation of dimerization-incapable CXCL8 mutants, which include deletion of C-terminal residues involved in α-helix formation and mutation of the residues within one of the β- strand.
Wild-type CXCL8 is first expressed as a 77 amino acid precursor and is then processed to its fully active 72 residues form (1-72, herein referred to as wild-type CXCL8)7.
1 Parts of this chapter were reprinted with permission from (1) Kendrick AA, Holliday MJ, Isern NG, Zhang F, Camilloni C, Huynh C, Vendruscolo M, Armstrong G, and Eisenmesser EZ The dynamics of interleukin-8 and its interaction with human CXC receptor I peptide. Protein science : a publication of the Protein Society 23: 464-480 (2014).
1
Multiple studies have revealed that the precursor form of CXCL8 is less active, supporting the biological importance of studying the mature, wild-type CXCL82, 8. In normal resting tissues CXCL8 is hardly detectable, but a stress response, which leads to an increase in inflammation, results in a highly elevated CXCL8. The specific regulation of CXCL8 expression has been attributed to nuclear factor-κB (NFκB) and nuclear factor for IL6 expression (NF-IL6) mediated transcriptional activity2. CXCL8's major role is to regulate inflammatory responses via its ability to recruit neutrophils to the sites of injury9. Neutrophil chemotaxis and degranulation is typically associated with pro-inflammatory functions of
CXCL8, however, cancer-promoting activity of this chemokine has also been identified.
CXCL8 has been reported to play an important role in regulating the progression of prostate, ovarian and lung cancers10-12. The specific events under CXCL8 regulation include angiogenesis, metastasis, tumorgenesis and the ability of CXCL8 expressing cancer cells to confer chemotherapeutic resistance2, 13. In cancer cells, CXCL8 can activate pro-tumorgenic pathways that involve endothelial growth factor receptor (EGFR)10, β-catenin14 and signal transducer and activator of transcription (STAT3)2 mediated signaling.
CXCL8 functional activity is often linked to an autocrine signaling mechanism. In particular; secretion of CXCL8 from cancer cells can enhance its signaling on neighboring cells, and hence lead to increase in cell proliferation, invasion and metastasis9, 10, 12. One important aspect of CXCL8 signaling is its ability to promote angiogenesis, an important hallmark of almost all cancers. Angiogenesis is an aptitude of cancer cells to promote growth of blood vessels that then supply necessary nutrients for tumor cell proliferation. In non-small lung cancer, secreted CXCL8 stimulates endothelial cells to increase tumor vasculature and promote angiogenesis10. Studies have also shown that CXCL8 involvement in promoting angiogenesis is linked to a vascular endothelial growth factor receptor
(VEGFR) mediated signaling and is linked to the intracellular interaction between VEGFR and CXCL8 receptors11.
2
CXCL8 signaling is initiated via its interaction with two G-protein coupled receptors,
CXCR1 and CXCR2, resulting in a diverse array of signaling events linked to the regulation of tumor microenvironment in several cancers. CXCR1 is targeted by CXCL8 and granulocyte chemotactic protein-2 (GCP-2) only, while CXCR2 is a more promiscuous receptor and it interacts with several other ligands along with CXCL8. These two receptors are homologous, with major sequence differences present in the extracellular N-terminal region implicated in ligand binding, thus explaining the specificity for their particular ligands.
The interaction between CXCL8 and CXCR1 has been described as a two-site process where site I comprises the initial interaction between N-terminal regions of both the CXCL8 and CXCR1 receptor, while site II involves the CXCL8 N-terminal (ELR motif) and the receptor distant extracellular region7, 14, 15. A recent elegant biophysical study performed in lipid bilayers using human full-length CXCR1 and CXCL8 did not fully support such interactions, suggesting that the binding process between CXCL8 and human CXCR1 appears to be mainly associated with the N-terminal region of the receptor16. Furthermore, using biophysical studies, Ravindran et al. proposed that upon receptor engagement, wild- type CXCL8 dissociates to form a CXCL8 monomer/receptor complex17. In contrast with these results, CXCL8 dimerization has been shown to be critical for neutrophil recruitment, one of the major CXCL8 roles in regulating inflammatory responses. These discrepancies may arise from the observation that most biophysical studies to date have been conducted with human CXCL8 and a peptide corresponding to the residues within the N-terminal region of a rabbit CXCR1 homolog, as opposed to its human counterpart, while biological studies reflect the interaction with the human CXCR1 (hCXCR1) sequence.
The dissociation constant (Kd) for CXCL8 dimer formation is reported to be in a 10-
20 µM range7, 17. As CXCL8 is secreted at high concentrations from injured or cancerous tissues, its local concentration can vary significantly, leading to the existence of both the monomeric and dimeric forms at different spatial and temporal occurrences. This differential
3
distribution of the CXCL8 dimeric and monomeric forms may suggest an important role for each of them in the functional activity of this chemokine. For example, the formation of
CXCL8 dimers at high concentrations may diminish the binding affinity of CXCL8 to its receptors, and thereby serve as a control mechanism to downregulate signaling.
Biomolecular NMR spectroscopy
Proteins are inherently flexible molecules, which undergo motions on multiple time scales. Characterizing the dynamics of a protein can help understand how the conformational changes are coupled it its function. Biomolecular nuclear magnetic resonance (NMR) spectroscopy is a biophysical technique that allows for identification of biomolecule interactions, selected reaction kinetics and dynamic motions on multiple biologically relevant timescales. This technique is based on a physical phenomenon in which nuclei with non-zero spin exhibit magnetic and angular momentum when positioned in a magnetic field. The most abundant NMR active nucleus in nature is hydrogen isotope (1H), however, biomolecular NMR often requires enrichment of nuclei within the molecules to obtain higher signal. The naturally abundant carbon (12C) and nitrogen (14N), the main building blocks of biomolecules, are not suitable for NMR because their spin numbers are equal to 0 and 1, respectively. Spin numbers equal to zero are not magnetically active and nuclei with a spin of 1 exhibit quadrupole relaxation, which leads to a short life span and subsequent line broadening or lack of signal. Therefore, NMR often relies on isotopic enrichment of biomolecules with 13C and 15N, which both have 1/2 spin numbers.
In the presence of a strong external magnetic field B0, a nucleus with spin 1/2 preferentially aligns with B0, which leads to a net alignment of bulk magnetization (M) of sample along the magnetic field (+z). A short radio frequency (RF) pulse can then be applied to generate a weaker magnetic field perpendicular to B0 (x or y) and to rotate M into a transverse plane (x-y). This leads to a procession of magnetization about B0 with a
4
characteristic Larmor frequency defined by the atom and B0. The nuclei process at specific rates that differ from one another due to small perturbations in their respective magnetic field. These perturbations are influenced by local environment changes due to protein structure, solvent exchange with charged residues, bond torsion and hydrogen bonding. The signal, time dependent current induced by nuclear procession after RF pulse, generated this way is measured in a form of a free induction decay (FID). FID is then converted into a frequency domain using Fourier transformation (FT). This transformation allows for the detection of three distinct properties: chemical shift (ω), signal intensity (I) and linewidth (λ).
Chemical shift reflects the position of the signal in the frequency spectrum and therefore, this property reports on the local chemical environment of the nuclei. The information obtained from ω can be used to identify local secondary structure properties of a biomolecule. Peak area, identified as height of the peak or volume under the peak, allows for identification of total number of atoms processing at a given frequency. Linewidth, measured as full peak width at half maximum height, correlates with the dynamic properties of a biomolecule in solution. Linewidth is directly related to transverse relaxation (R2 = λπ) and allows for the identification of dynamic changes within the pico-microsecond (ps-micros) timescales.
The three above mentioned parameters can be altered by chemical exchange, a dynamic process that exposes an NMR nucleus (NMR probe) to distinct chemical states
(environments) in a time dependent manner. This process can be due to an exchange between free/bound ligand state, dimer/monomer equilibrium or other conformational changes. Although the described study focuses on a two-state system it is important to mention that other more complicated systems, where an NMR probe samples multiple states, can likely exist. A two-state system can be described by population occupancies (PA and PB), the rate of exchange (kex = kAB + kBA) and the chemical shift difference between the two states (Δω). The relationship between kex and Δω separates the NMR spectra into three
5
distinct regimes (Figure 1). In fast exchange (kex >> Δω), a single peak occurs and represents population-averaged chemical shift position. Slow exchange (kex << Δω) allows for detection of two distinct populations for each chemical environment as long as the population occupancy of those states is high enough. Intermediate exchange (kex ~ Δω) included all other scenarios and leads to significant peak linebroadening.
NMR allows for studying protein-ligand binding and the determination of a dissociation constant (Kd) in the case of fast or intermediate exchange processes. With the use of heteronuclear single quantum coherence (HSQC) experiment weak binding interactions can be assessed in a quantitative way. Upon addition of increasing concentrations of ligand to a constant concentration of a protein of interest and measuring the extent of chemical shift perturbation changes, a binding isotherm can be calculated (see
-5 Methods for details). This calculation is valid for weak interactions (Kd > 10 M) when the binding event is exhibiting fast exchange. In the case of intermediate exchange only an estimation of dissociation constant can be typically made due to low resolution and high signal-to-noise ratio. High-affinity ligand binding is due to slow ligand dissociation and, hence, slow exchange, which does not allow for enough observable chemical shift perturbations to obtain a binding isotherm.
NMR is an excellent method to study ps-ns scale timescale backbone and side-chain fluctuations as well as millisecond (ms) conformational rearrangements. Backbone and side chain dynamics can be measured by the use of longitudinal (R1) and transverse (R2) relaxation rates. R1 and R2 can be measured by surveying the decay of magnetization as a function of time delay when magnetization is stored in the longitudinal axis or transverse plane, respectively. The spectral density, J(ω), which describes the amplitudes of motion experienced at a given frequency is directly related to the relaxation rates. J(ω) is also influenced by the global tumbling time of the biomolecule (ns motion) in addition to any localized bond motions (ns-ps motion). The interpretation of relaxation motion is commonly
6
done by the use of Lipari-Szabo model free approach18. This approach assumes that internal motions of a biomolecule are much faster than its global motions but otherwise no additional structure related motion assumptions are made, hence the "model free" name.
The parameters used to describe this model are: global tumbling time (τe), site-specific
2 correlation time (τm), site-specific order parameter (S ), and µs-ms exchange (Rex). R1 and
R2 relaxation rates can be interpreted across the sequence for each observable residue and such observation reflect on regional internal motions within the protein.
Slower timescale motions (ms time scale) can be monitored by the use of Carr-
Purcell-Meiboom-Gill relaxation dispersion experiment (CPMG) also called R2 relaxation dispersion (RD)19. CPMG-RD allows for more quantitative measurement of dynamic motions and quantitative evaluation of specific population states. This technique detects motions due to side chain reorientation, loop motions and other structural rearrangements including dimerization propensities of a particular region of a biomolecule. The basic CPMG experiment involves a continuous relaxation time during which a 180o refocusing pulses with different frequencies are applied in the transverse plane to allow for magnetization decay through the R2 relaxation. The R2 relaxation measured in CMPG-RD experiment includes information from the background relaxation level (R20) and any additional contribution from chemical exchange on the µs-ms timescale (Rex, R2 = R20 + Rex). The R2 relaxation obtained from CMPG-RD experiment is then described by two-state Carve-Richards equations20.
Although CPMG-RD can be used to monitor system in all three exchange regimes, information about rates of inter-conversions (kex), relative populations (PA) and the absolute values of chemical shift differences between the exchange states (|Δω|) can only be obtained for intermediate exchange regime.
For the processes undergoing ms motions, Δω obtained from CPMG-RD experiments does not reflect the sign of chemical shift difference. This information can be
7
obtained from comparing chemical shifts in the indirect dimension from HSQC and HMQC
(heteronuclear multiple-quantum coherence) experiments collected on the same sample in the same magnetic field. This combination of two standard NMR experiments (herein termed as HS/MQC) takes advantage of the fact that the residues undergoing chemical exchange on the µs-ms timescale give rise to small chemical shift changes between these two spectra21, 22.
Study rationale
Several studies have portrayed the importance of dimerization on the activity of
CXCL8 in vitro and in vivo however, the molecular and functional consequences of such dimerization are not fully characterized. The described here study aims to characterize the specific mechanism of CXCL8 dimer and monomer interaction with human CXCR1 receptor peptide. Previous reports portray conflicting results in regard to dimer dissociation upon
CXCR1 receptor engagement and, thus, this study attempts to explain the molecular details of this interaction. Furthermore, the interaction between CXCL8 and its receptor has mainly been described in regard to the rabbit homologue or a synthetic peptide corresponding to the N-terminus of CXCL8. The described here research addresses the specifics of CXCL8 interaction in regard to the biologically relevant human CXCR1 peptide.
Results
CXCL8 monomer and dimer binds hCXCR1pep with slightly different affinities
In order to test the binding of human CXCL8 to its human target receptor, we expressed and purified a peptide corresponding to the N-terminal region (residues 9-29) of human CXCR1 (hCXCR1pep, Figure 2). We made the selection of the length of the peptide based on previous studies that attempted to characterize the interaction between CXCL8 and hCXCR1 in which peptides spanning residues 1-40 of the hCXCR1 N-terminal region or a shorter construct were utlilized7, 23. Comparable binding characteristics were reported with
8
either a truncated shorter peptide (residues 9-29) or a truncated peptidomimetic (residues
15-19 replaced by a chemical linker). The initial CXCL8 construct contained a 6xHis tag and a thrombin cleavage site that generated a final construct with an N-terminal overhang after thrombin cleavage (Figure 2). We also generated an additional monomeric CXCL8 construct in order to test the contribution of dimerization to the binding process. The dimerization- incapable mutant form of CXCL8 (called CXCL8M, "M" standing for monomer) was constructed by performing site-directed mutagenesis within the dimerization domain as previously described (Figure 2)24. While both wild-type and monomeric constructs were initially generated with an N-terminal overhang resulting from thrombin cleavage (herein called CXCL8Synthetic and CXCL8MSynthetic, respectively), these constructs were also produced using a Factor Xa cleavage site that allowed for complete removal of the purification tag and the cleavage site hence leaving no overhang. This approach permitted to also test the biologically mature forms of CXCL8 without any overhang present (herein called CXCL8 and
CXCL8M). We tested each construct using NMR and size-exclusion chromatography
(Figure 3) to identify their in-solution behavior and concluded that both constructs are well folded and behave as a dimer and monomer, as expected. Comparative studies between the synthetic and the biological form have allowed to account for the importance of the
CXCL8 N-terminus in the binding interaction (Figure 2). In order to assess the binding interface and affinity we utilized 1H,15N-HSQC solution NMR experiments to monitor chemical shift perturbations upon ligand addition. Unlabeled hCXCR1pep was titrated into the multiple 15N-labeled CXCL8 constructs, and binding-induced chemical shift perturbations were monitored upon peptide engagement.
We sought to specifically test the sensitivity of CXCL8/hCXCR1pep binding to different buffer conditions since previously reported binding affinities vary significantly. Such discrepancies may have arisen from the various conditions and/or specific constructs used7,
8, 17, 23, 25. We performed a more comprehensive screen of the different conditions by
9
executing all experiments in the presence of some of the previously reported buffers and using different CXCL8 constructs. In particular, we performed initial titrations in the presence of phosphate buffer at pH 6.5 for synthetic CXCL8 constructs (Figure 4). Although both synthetic CXCL8 constructs engaged hCXCR1pep, the monomer exhibited slightly tighter binding affinity as determined by chemical shift perturbations and the associated binding isotherms with dissociation constants of 544±25 µM for CXCL8Synthetic and 440±30 µM for
CXCL8MSynthetic (Table 1 and Figure 4).
Since these interactions were relatively weak, we next sought to determine whether they were specific. The contribution of a single Asp13 residue within the hCXCR1pep to binding was surmised based on reverse titrations where we added unlabeled CXCL8Synthetic into 15N-labeled hCXCR1pep that we previously assigned (Figure 5). We observed the largest chemical shift changes in Cα and Cβ for Asp13 of hCXCR1pep upon binding CXCL8
(Figure 5), providing initial evidence as to the electrostatic nature of this interaction.
Consistent with this conclusion, an Asp13!Ala mutant hCXCR1pep showed marked reduction in binding affinity to CXCL8Synthetic with a dissociation constant of 1.3±0.05 mM
(Figure 5). Interestingly, while there have been mutagenesis studies of the hCXCR1 N- terminal region, Asp13 has not been previously characterised as important for binding 26 and thus, in addition to revealing specificity, our studies have identified a primary contribution to the CXCL8/CXCR1 interaction. Noteworthy, we obtained similar affinities upon using the wild-type CXCL8 constructs (Table 1 and Figure 4), with 550±26 µM and 460±27 µM for
CXCL8 and CXCL8M, respectively. Consistent with the conclusion that electrostatics play an important role in the CXCL8/hCXCR1pep interaction, we detected a significant increase in binding affinities when titrations were performed in the presence of HEPES buffer instead of phosphate buffer (Figure 4). Unfortunately, we were not able to confidently determine the binding affinity for wild-type constructs in the HEPES buffer due to slight sample precipitation that was especially evident for CXCL8M. However, a comparison of the
10
chemical shift perturbations at the first few titration points, prior to precipitation, revealed significantly larger shifts in HEPES buffer, relative to the ones collected in phosphate buffer, potentially eluding to a higher binding affinity. This observation is in agreement with other studies in HEPES buffer that reported a dissociation constant of 10 µM using Isothermal
Titration Calorimetry27. Although both HEPES and phosphate buffers are routinely utilized to mimic physiological conditions, the actual biological environment is more complicated because of the presence of a variety of other molecular species. Hence, the binding affinities measured in those buffers are likely to be influenced by other factors (e.g. other ions or proteins). Based on the NMR titrations data we concludes that the interaction between CXCL8 and CXCR1pep involves a number of charged residues, which are described below, suggesting that the nature of this interaction is primarily electrostatically driven. This electrostatic interaction would likely be masked by high ionic strength buffers
(i.e. phosphate buffer), consistent with the aforementioned results. Based on the NMR results we conclude that: (i) the CXCL8 monomer binds hCXCR1pep with a slightly higher affinity than the CXCL8 dimer, (ii) modifications to the N-terminal region of CXCL8 bear little consequence to binding, and (iii) the CXCL8/hCXCR1pep interaction is primarily electrostatic.
The peak widths observed in the 1H,15N NMR spectra during titrations suggest that the exchange process is in a fast NMR time regime, which is consistent with the relatively weak binding constants determined above. Based on the previously published crystal structure of CXCL84, we can deduce that residues exhibiting the largest chemical shift perturbations are located in the N-terminal loop (Thr12, Phe17, His18, Phe21), the turn preceding β3-strand (Ser44, Asp45), β3-strand (Glu48, Leu49, Cys50) and the C-terminal α- helix (Val61 and Val62). A mapping of these residues onto the crystal structure revealed that they are highly localized (Figure 4 and Figure 6). With the exception of residues Asp45 and
Val62, these results are in agreement with previously published studies describing similar
11
regions within CXCL8 for the interaction with the longer or modified hCXCR1 peptide7, 23.
The binding interface changed only marginally for the monomeric form of CXCL8 (Figure 6), with the involvement of additional residues exhibiting perturbations within the first β-strand
(Tyr25 and Arg26) and the α-helical region (Lys67 and Arg68). We did not observe any significant chemical shift perturbations for the N-terminal region (ELR motif) for all CXCL8 constructs, suggesting that these residues are not critical for the N-terminal hCXCR1 peptide interaction. This lack of the involvement of the CXCL8 ELR motif in binding is consistent both with our results that detected similar affinities for the synthetic and wild-type
N-termini of CXCL8 above, as well as with several previous studies16, 28.
CXCL8 dimers do not dissociate upon hCXCR1pep engagement
CXCL8 dimer dissociation has been proposed to be a prerequisite for binding peptides derived from CXCR127, 29. The nature of the CXCL8 interaction with the hCXCR1 N- terminal region has been extensively characterized for the rabbit homolog15, 30, but there are limited data available describing the interaction of human CXCL8 with its associated hCXCR1. Thus, to establish whether CXCL8 dimers dissociate upon binding to hCXCR1pep as previously proposed, we determined the correlation times (τc) of CXCL8 in its free and bound states, as this parameter is sensitive to molecular weight31. We collected R1
(longitudinal) and R2 (transverse) relaxation rates for backbone amides at 600 MHz, for both wild-type dimeric and mutated monomeric constructs in both free and hCXCR1pep bound states. Based on the dissociation constants described above, complete saturation was not possible in phosphate buffer but only in HEPES buffer. Thus, data are presented here in
HEPES buffer (Figure 7). However, comparable results were found in phosphate buffer as well (see Figure 11 for the free state). Under the sample conditions reported here, the average R1 relaxation rates for the free wild-type CXCL8 and CXCL8M constructs were approximately 1.4 s-1 and 2 s-1, respectively (Figure 7), while in the bound states these
12
averages change to 1 s-1 and 1.6 s-1, respectively (Figure 7). The average R2 relaxation rates were approximately 12 s-1, and 8 s-1 for the free wild-type CXCL8 and CXCL8M constructs, respectively (Figure 7), while the bound states averages increased to 21 s-1 and
13 s-1, respectively (Figure 7). Based on the measured R2/R1 ratios, we calculated correlation times for the free and bound proteins using the equations described by Larsson et al.31.
For free CXCL8 and CXCL8M, the calculated correlation times were 8.7 and 5.7 ns, respectively, and 14.9 and 8.9 ns for the bound counterparts, respectively (Figure 8 insert).
In folded proteins, the effective correlation time is a function of the molecular weight of the protein up to 25-30 kDa32, 33. Thus, the evaluation of correlation times allows for relatively accurate estimation of protein molecular weights31, 34. Therefore, this method can readily be used to monitor protein dimerization and protein interactions in general, based on changes that alter molecular weight35, 36. Using this established dependence of correlation time to molecular weight, we compared experimental correlation times to a standard curve obtained from published correlation rates for known proteins (Figure 8, black dots and Table 2), which were acquired at the same temperature32, 33. From these measurements, we estimated the molecular weight of free CXCL8 and CXCL8M to be 14.7 kDa and 9.4 kDa, respectively
(Figure 8, blue dots). These values estimated from experimental data correspond well to the molecular weights, calculated from the structure of the wild-type CXCL8 dimer (17 kDa) and
CXCL8M monomeric form (8.6 kDa). We then extended those estimations to the relaxation data measured in the bound species providing molecular weights of 25.1 kDa and 15.2 kDa for the CXCL8 and CXCL8M complexes with hCXCR1pep, respectively (Figure 8, green dots). Considering that the wild-type CXCL8 dimer and hCXCR1pep are 17 kDa and 2.9 kDa, respectively, the molecular weight of CXCL8 dimer bound to two molecules of hCXCR1pep is 22.8 kD (i.e., 17 kDA + 2*(2.9 kDA)). This value agrees very well with the molecular weight estimated from our experimental data (25.1 kDa), which indicates that the
13
dimer does not dissociate upon binding to the hCXCR1pep. Calculated values of correlation times for both CXCL8 and CXCL8M complexes were slightly higher than their actual molecular weights, which was likely due to the extended hCXCR1pep bound unstructured regions. However, the calculated wild-type CXCL8 dimer complex lies well within the bounds of an intact dimer complex, and is much larger than that of the CXCL8 monomer complex (Figure 8, green dots). We also obtained similar values for the CXCL8 and
CXCL8M constructs in phosphate buffer (Table 3). Collectively, these data indicate that
CXCL8 does not dissociate upon receptor peptide engagement and, furthermore, the
CXCL8 dimer stays intact with two molecules of hCXCR1pep bound per CXCL8 dimer.
These findings are consistent with previous studies performed with another hCXCR1 peptide, but not with some of the conclusions utilizing a rabbit CXCR1 peptide that indicated dimer dissociation upon rabbit peptide binding17, 27, 28.
CXCL8 exhibits inherent flexibility within the binding region
Protein flexibility has been shown to play a critical role in enzyme-substrate and protein-ligand interactions, with binding sites and active sites often exhibiting inherent flexibility over multiple time regimes37-39. In order to determine whether a ligand such as
CXCL8 may also be inherently flexible within regions important for function, i.e., receptor engagement, we probed the dynamics on multiple time scales for CXCL8. R1 and R2 relaxation rates described above that allowed us to probe the molecular weight of CXCL8, both free and bound, can also be used to evaluate local mobility. For proteins that are in general outside of the “extreme narrowing limit” (i.e., τc>2.65 ns from 1/(2πω), with ω=60.8
MHz for 15N on a 600 MHz spectrometer), relatively high R1 relaxation rates, together with low R2 relaxation rates, identify regions that exhibit a high degree of mobility, which indicates locally unstructured regions. This high mobility is observed for several regions that include the N- and C-termini of the intact CXCL8 dimer (Figure 7 and Figure ). Additional
14
residues that exhibit R1 relaxation rates higher than a standard deviation above the mean are Ile11, Thr12, His33, Ala35, Asp45 and Trp57 (Figure 7). His33 and Ala35 surround
Cys34, which is located in a loop region and forms a disulfide bond with Cys7. The high mobility detected for these residues is likely due to the position of those residues in the flexible loop. Interestingly, Thr12 and Asp45 are both directly involved in binding hCXCR1pep. Furthermore, all the residues exhibiting elevated relaxation rates are located in proximity to the binding interface. Thus, the increased flexibility of these amino acids appears to support the notion that residues may be flexible for function, which in this case is receptor binding. Finally, it should be noted that both the N- and C-termini of the monomeric
CXCL8 mutant exhibit both low R1 and low R2 relaxation rates (Figure 7, right panel) because these regions are so flexible that they actually lie within the extreme narrowing limit with calculated local correlation times lower than 2.65 ns.
In order to identify µs-ms motions, we also collected HSQC and HSMC spectra at the same magnetic field. The residues exhibiting elevated H(S/M)QC exchange induced shifts (one standard deviation above average) in the wild-type CXCL8 are Ala15, Arg26,
Ile40, Cys50, Val58, Asn59, Arg60 (Figure 9), with Ala15, Arg26 and Cys50, exhibited the highest changes. The identified residues are mainly located in the proximity of the binding site as shown in Figure 9. These results suggest once again that inherent flexibility may play an important functional role in the binding interface of CXCL8 on multiple timescales.
Furthermore, we also collected 15N-R2-CPMG experiments to identify motions within the millisecond timescale40. However, we did not detect any measurable 15N-R2-CPMG relaxation dispersion (Figure 9), suggesting that slow dynamics within wild-type CXCL8 are primarily confined to the µs timescale.
15
CXCL8M dimerizes at high concentrations
Although the engineered CXCL8M mutant form has previously been described as a strictly monomeric species, we have found that it has a tendency to dimerize24. Specifically, as opposed to wild-type CXCL8, we observed detectable exchange in the ms regime for 14 out of 72 residues within CXCL8M as shown by 15N-R2-CPMG dispersions that was concentration dependent (Figure dark green versus blue lines). Both this concentration dependence and the fact that the regions undergoing exchange primarily localized to the dimerization interface strongly suggest that CXCL8M is capable of dimerizing (Figure ).
Moreover, all residues within CXCL8M exhibited similar dispersion profiles at the same concentration, which suggests a global cooperative process (i.e., dimerization). Utilizing dispersion data from two static magnetic fields (600 MHz and 900 MHz in Figure , solid vs. dashed lines) allowed for the determination of the exchange rate between dimer/monomer
(130±30 s-1) as well as the population of the dimer (7% dimer) at the 1 mM concentration used for the CXCL8M. Residues exhibiting exchange include the first β-strand, residues within the α-helix most proximal to the C-terminus and residues adjacent to the α-helix but positioned at the end of 2nd and 3rd β-strands (Figure ). Thus, caution should be taken in interpreting the R2 relaxation rates for CXCL8M, since several residues have contributions from exchange to the dimer. Nonetheless, despite the mutations to the dimer interface that induce dimer dissociation, our results clearly show that CXCL8M does dimerize weakly within the millimolar concentrations utilized for NMR, suggesting that the association constant is within this range.
Chemical shift-based calculations identify fast time scale dynamics
Chemical shifts have long been used to predict protein structural propensities41.
More recently, it has been shown that it is possible to increase the amount of information that can be extracted from them by introducing methods to translate chemical shifts directly
16
into protein structures42, 43 and protein structural fluctuations44, 45. This latter approach provides a complementary route to the use of relaxation data to study protein dynamics. In this context, the use of unrestrained molecular dynamics simulations provides a powerful tool for interpreting relaxation data, as shown recently by Skrynnikov et al46. However, since this type of analysis requires millisecond long trajectories, we (in collaboration with Michele
Vendruscolo and Carlo Camilloni) analyzed the equilibrium fluctuations of CXCL8 and
CXCL8M using molecular dynamics simulations with chemical shift restraints. In the following, we first used chemical shifts to predict the secondary structure propensities and then we performed chemical shift restrained MD to analyze the conformational fluctuations within the predicted ensembles.
The differential binding between CXCL8 and CXCL8M suggests that there may be an underlying difference in the structure and/or dynamics between the two forms. Thus, we analyzed our assigned carbon chemical shifts using the δ2D method44, 45, which translates the chemical shifts into the secondary structure populations. The comparison of the results for free CXCL8 (Figure 11 – right panel) and free CXCL8M (Figure 11 – left panel) indicate that the C-terminal α-helix is less well formed in the mutant, and that there is an overall increase in fluctuations in the regions corresponding to the β-sheet. Interestingly, the N- terminal region of CXCL8M appears to be slightly less flexible as compared to CXCL8, while there is increased flexibility in CXCL8M in two loops comprising residues 28-31 and 50-53
(Figure 11). These data, for both CXCL8 and CXCL8M, are consistent with the relaxation data that identified similar flexible regions (Figure 7). To gain further insight into these structural differences, we utilized our chemical shift data for both CXCL8 and CXCL8M in molecular dynamics simulations employing chemical shifts as replica-averaged restraints47.
Since these chemical shifts were initially assigned in phosphate buffer, we also collected relaxation data in this same buffer (Figure 11), which were similar to those relaxation rates
17
collected in HEPES buffer presented above (Figure 7). Interestingly, local RMSFs calculated for both the dimeric and monomeric forms are largely mirrored by the R1 relaxation rate
(Figure 11, solid black vs. solid red lines), suggesting that the calculated ensembles here are largely descriptive of the fast time scale dynamics. A possible reason that the R2 relaxation rates may not match as well is that they may have contributions from slower motions that are not necessarily captured within the chemical shift-based calculations. As for the fundamental difference between the dimer and monomer, the increased disorder within the C-terminal region and two loops comprising residues 28-31 and 50-53 can readily be visualized with a partial unfolding of the C-terminal α-helix in CXCL8M relative to the wild- type CXCL8 dimer (Figure 11, bottom panels). These results are illustrated particularly clearly by comparing the free energy landscapes of CXCL8 and CXCL8M as a function of the RMSD of the C-terminal α-helix and the RMSD of the two loop regions (Figure 11).
Discussion
In this study we have investigated the binding of CXCL8 to a peptide corresponding to the N-terminal region of human CXCR1. Our results support previous NMR solution findings in which the interaction between CXCL8 and hCXCR1 appears to be weak with micromolar dissociation constants23. However, our study is the first to utilize NMR relaxation data to show that, upon engagment of the human receptor peptide, the dimer does not dissociate as previously proposed17, 27. These results suggest that at physiological concentrations, where both the CXCL8 dimer and monomer likely exist, both forms may initiate the interaction with the receptor, and thus be important for biological functions.
Biological data indicate that in some cases the CXCL8 dimer is more active than the monomer, but studies showing the opposite reults have also been reported. For example, several biological studies demonstrate that dimerization deffective CXCL8 mutants (i.e., monomeric CXCL8 at experimental concentrations) are less active in colony formation in
18
myeloid progenitor cells and inhibition of TNF-α stimulated peroxide induction in neutrophils24, 48, 49. Contrary to this, at low concentrations a trapped CXCL8 monomer was reported to be more active than the wild-type CXCL8 dimer in a neutrophil migration assay50.
Our biophysical studies indicate that a mutational variant of CXCL8 designed to remain monomeric (CXCL8M) binds only slightly more tightly to hCXCR1pep (Figure 4 and Table
1). However, a recent study in lipid bilayers has shown that CXCL8 binds much tighter to the
N-terminal regions of CXCR1 in these conditions than in the absence of lipids16. This result may warrant further comparitive studies between CXCL8 and CXCL8 monomeric mutants to discern if this difference in affinity holds in the context of lipids.
One of the most remarkable findings in the field of protein dynamics has been that many enzymes are inherently flexible within the regions required for binding. For example, enzymes like cyclophilin-A22, 37, dyhydrofolate reductase51, and RNAse A52 exhibit conformational dynamics within their catalytic sites that are critical for catalytic turnover.
However, such inherently flexible regions are likely ubquitiously present within non-catalytic proteins as well, such as cytokines that target multiple receptors which may also be flexible in order to allow for differential engagement. A case in point is interleukin-3 (IL-3), which has previously been shown to exhibit dynamics on multiple timescales which has been suggested to allow for binding promiscuity to multiple receptor subunits6. Our data presented here reveal a similar trend with CXCL8. Specifically, we show evidence that there may be inherent dynamics localized to the hCXCR1pep binding region (Figure 9). These motions include ps-ns dynamics identified through R1/R2 relaxation rates (Figure 7 and Figure 11).
Additionally, the presence of elevated H(S/M)QC exchange induced shifts with no observable R2-CPMG dispersion in the CXCL8 dimer indicates that slower motions are likely confined to the ms timescale (Figure 7). Thus, despite the fact that IL-3 and CXCL8 exibit no structural similarity with one being a four-helix bundle and the other a mixed a/b structure, respectively, both interleukins are dynamic molecules with increased flexibility
19
largely present within their receptor binding sites. These results support the view that proteins, such as enzymes and ligands, have evolutionarily evolved to have a flexibility compatible with their function37. This view also implies that proteins can undergo conformational selection in order to target their binding partners39.
Chemical shift-based calculations have recently become a powerful method to probe structure and dynamics of proteins and here we utilized such methods to compare the structural integrity of wild-type CXCL8 dimers with that of the mutated CXCL8 monomers.
Initial clues to the structural differences between the CXCL8 dimer and monomer were observed in both the relaxation data (Figure 7 and Figure 9) and RMSF (Figure 11).
Chemical shift-based calculations revealed an unfolding of the C-terminal region of the
CXCL8 monomer (Figure 11), which is involved in stabilization of the CXCL8 dimer. Such flexibility in the CXCL8M may also explain the slightly tighter binding affinity relative to the
CXCL8 dimer. This result likely leads to more interactions between the C-terminal α-helix of the CXCL8 monomer with hCXCR1pep that is supported by increased chemical shift perturbations (Figure 6). Interestingly, the local RMSFs for both the CXCL8 dimer and monomer were highly predictive of the R1 relaxation rates, suggesting a link between these ensembles and fast dynamics (Figure 11). Finally, the data presented here directly shows that the mutated form of CXCL8 (i.e., CXCL8M) dimerizes at high concentrations.
Consistent with previous findings24, this CXCL8 mutant form is predominantly a monomeric species as indicated by size exclusion chromatography (Figure 3) and also via correlation time measurements (Figure 7). However, 15N-R2-CPMG relaxation dispersion data provided an additional method to probe for self-association and revealed a concentration-dependent self-association through the dimer interface (Figure ). Furthermore, the restrained molecular dynamics simulation data supports the 15N-R2-CPMG relaxation dispersion results by showing that the most energetically favorable conformations sampled by CXCL8M are highly similar to those sampled by wild-type CXCL8 (Figure 11). Although, overall CXCL8M
20
samples a much broader free energy landscape than its wild-type counterpart. This similarity in location of the lowest part of the free energy landscape provides further evidence that
CXCL8M is capable of dimerizing.
This is an important finding since several different CXCL8 mutants are commonly used for biological and biophysical studies. Thus, our data shows that caution must be taken when evaluating the results of such studies, as dimerization still persists at higher concentrations. Although such a dimerization may not initially appear to be a problem at the lower concentrations used for assaying CXCL8 biological activities, elevated local concentrations may give rise to dimer formation and, thus, should be recognized as a potential complicating factor.
In conclusion, we have shown through NMR solution studies that CXCL8 dimers do not dissociate upon binding to the N-terminal region of human CXCR1 and that CXCL8 appears to be inherently dynamic within its receptor-binding region. Moreover, we have described how the differences in the previously reported binding affinities of CXCL8 to
CXCR1 arise at least in part from the electrostatic nature of this interaction, which was studied here by comparative measurements in both phosphate and HEPES buffer.
Future directions
The results described herein have important implications in the design of therapeutics that may specifically target CXCL8. Our studies suggest that both the dimeric and monomeric forms of CXCL8 will likely have to be targeted simultaneously in order to block the pro-inflammatory activities of this chemokine. Our discoveries along with others suggest that the binding of CXCL8 to CXCR1 involves other interfaces outside of the N- terminal CXCR1pep region. Further studies are now required to establish those interactions and the possible contribution of other regions of CXCR1 to the interaction interface. It appears that a system, which utilizes lipid bi-layer should be applied in here to determine the
21
specific residues involved in the interaction. Since CXCR1 is a multi-pass transmembrane protein and is inherently embedded in cellular membrane it seems necessary to study the interaction between CXCL8 and this receptor in context of lipid environment. Once the specifics of this interaction are characterized a D-peptide could be designed to block the interaction and, hence, inhibit the pro-cancerous activity of CXCL8. In addition a different peptide could be designed to target the monomeric vs. dimeric form of CXCL8 and inhibit the activities that differ between those two forms.
22
Table 1 Dissociation constants obtained from chemical shift perturbation for the different CXCL8 constructs
Kd [µM] Constructs and buffer condition Dimer Monomer CXCL8Synthetic : hCXCR1pep, 50 mM phosphate, 544 ± 25 440 ± 30 150 mM NaCl, pH 6.5 CXCL8 : hCXCR1pep, 50 mM phosphate, 150 550 ± 26 460 ± 27 mM NaCl, pH 6.5 CXCL8 : hCXCR1pep, 50 mM HEPES, 50 mM n/a n/a NaCl, pH 7.0
Table 2 The proteins, their molecular weight and correlation time used for generation of graph in Figure 6.
Swissprot MW Correlation Protein namea ID# (kDa) time (ns)
Lipoprotein Q883K7 7.2 5.1 Ubiquitin P62979 9 4.4 50S ribosomal protein LX O27647 10.5 7 Interferon, alpha-inducible protein Q5SVA4 10.7 5.7 Cytochrome c-type biogenesis protein CcmE Q72D78 10.9 6.5 Protein of Unknown Function Q5E7H1 11.2 6.3 Replication protein A Q6LYF9 11.8 7.8 CD147 Ig2 domain53 P35613 12.3 9.6 Putative secretory antigen Q49ZM2 12.4 7.1 O6-methylguanine-DNA methyltransferase A6B4U8 12.5 8.1 Hypothetical Membrane Associated Protein Q812L6 13.1 8.8 Uncharacterized protein Q8U1U6 13.6 9 UPF0339 protein SO_388 Y3888 13.8 7.7 Villin 14T54 P02916 14 10.2 Lipoprotein spr precursor P0AFV4 15.8 10 Uncharacterized lipoprotein yajI precursor P46122 18.8 11.3 Protein of Unknown Function Q8TTH3 20.2 12.2 CD147 Ig0 domain (dimer)55 P35613 26.2 15.8 56 Maltose binding protein P02916 43 23 aAll correlation times were derived from Rossi et. al., except for when otherwise stated33.
23
Table 3 The extracted correlation times and corresponding molecular weights (MW) for the free and bound constructs from relaxation dispersion experiments acquired in the presence of phosphate buffer.
MW calculated MW estimated Correlation time* from structure from exp. (ns) (kDa) values** (kDa)
CXCL8Synthetic 9.7 17.6 16.4 CXCL8Syntheticbound 15.4 23.2 26 CXCL8 8.9 17 15 CXCL8bound 12.5 22.7 21.1 * Correlation time was calculated using R1 and R2 relaxation dispersion measurements collected at 25oC and 600 MHz. ** The MW was estimated from experimental data by fitting it into linear regression (τc = 0.59*MW) generated from previously published data (Figure 6 and Table 2). All data within this figure were collected in phosphate buffer at 25 oC at 600 MHz.
24
Figure 1 NMR chemical exchange regimes Lineshapes for two NMR probes undergoing two-state exchange. I the slow exchange regime two distinct peaks are observed. Both peaks are broadened and shifted toward the average in intermediate exchange. The peak in fast exchange appears in the average position with an intensity corresponding to the sum of both peaks.
25
Figure 2 Sequence of the CXCR1 N-terminal region, and sequence and structure of interleukin-8 (CXCL8) (A) Alignment of the amino acid sequence of the human CXCR1 (hCXCR1) N-terminal region with that of the rabbit CXCR1 (rCXCR1) N-terminal region. The peptide used in the study (hCXCR1pep) and the binding mutant (hCXCR1D13Apep) are also shown, with the post-purification overhang highlighted in green. (B) Amino acid sequences of human CXCL8 constructs utilized in the study. The post-purification overhang residues are highlighted in green and constructs containing such overhangs are described as synthetic. The double mutation that blocks CXCL8 dimerization and is thus a CXCL8 monomer (L25Y and V27R) is highlighted in red box. (C) Ribbon diagram of the CXCL8 dimer crystal structure. N- and C- termini of each monomer are labeled and the residues (L25 and V27) mutated to form monomer are mapped onto one of the monomers. PDB accession code 3IL8, Human CXCL8 GenBank accession code 3576, hCXCR1 GenBank accession code L19591, rCXCR1 GenBank accession code M58021.
26
Figure 3 1H,15N HSQC NMR spectra of uniformly 15N labeled CXCL8 constructs utilized in the study (A) CXCL8Synthetic (black) and CXCL8MSynthetic (red) in 50 mM phosphate, 150 mM NaCl, pH 6.5; (B) CXCL8 (black) or CXCL8M (red) in 50 mM phosphate, 150 mM NaCl, pH 6.5; (C) CXCL8 (black) or CXCL8M (red) in 50 mM HEPES 50mM NaCl pH 7.0. All NMR data were collected at 25 oC at 900 MHz. (D) Representative size exclusion chromatogram (S75) of CXCL8 (black) and CXCLM8 (red) in phosphate buffer indicates the formation of a dimer and a monomer, respectively. The same distribution was observed in HEPES buffer and for CXCL8Synthetic and CXCL8MSynthetic constructs.
27
Figure 4 CXCL8 engages human CXCR1 peptide weakly with increased affinity in the monomeric CXCL8 form (A-C) Expanded regions of 1H/15N HSQC NMR titration experiments for the uniformly 15N labeled CXCL8 (left panels) and the monomeric CXCL8 mutant (right panels). Increased concentrations of unlabeled hCXCR1pep were titrated into 15N labeled protein (black - free protein). (A) CXCL8Synthetic or CXCL8MSynthetic in 50 mM phosphate, 150 mM NaCl, pH 6.5; (B) CXCL8 or CXCL8M in 50 mM phosphate, 150 mM NaCl, pH 6.5; (C) CXCL8 or CXCL8M in 50 mM HEPES, 50 mM NaCl, pH 7.0. The molar ratios of the unlabeled hCXCR1pep to 15N labeled protein were 0.2 (pink), 0.4 (green), 1.5 (blue), 4 (purple), 8 (red) in (A) and (B) and .25 (yellow), 0.5 (brown), 1 (light green), 2 (dark green) in (C). (D) Residues exhibiting normalized chemical change of 0.6 ppm or above � � (defined by (�����) + (�����) ) upon hCXCR1pep peptide engagement are painted red on the CXCL8 dimer and monomer structures. Protein Data Bank ID 3IL8. All data were collected at 25 oC at 900 MHz.
28
Figure 5 The conformation of specificity in CXCL8 binding to hCXCR1pep (A) 1H,15N HSQC NMR spectrum of uniformly 15N labeled hCXCR1pep (free – black, bound – red). Expanded regions of 1H,15N HSQC NMR titration with unlabeled wild-type CXCL8 is showed in the right top corner. The molar ratios of the unlabeled CXCL8 to 15N labeled hCXCR1pep were 0 (black), 0.5 (blue), 1 (yellow), 2 (green), 3 (pink) and 4 (red). (B) Cα and Cβ chemical shift perturbations of 1H,15N,13C hCXCR1pep upon wild-type CXCL8 engagement. (C) Expanded region of 1H/15N HSQC NMR titration experiments for the uniformly 15N labeled wild type human CXCL8. Increased concentrations of unlabeled hCXCR1D13Apep were titrated into 15N labeled protein. The molar ratios of the unlabeled hCXCR1pep to 15N labeled protein were 0 (black), 0.2 (pink), 0.4 (green), 1.5 (blue), 4 (purple), 8 (red). All data shown within this figure were collected in phosphate buffer at 25 oC at 900 MHz.
29
Figure 6 Changes in chemical shifts upon CXCL8 binding to hCXCR1pep Normalized ! ! 1 15 chemical change (defined by (5��!!) + (��!"!) ) per H/ N labeled protein residue upon hCXCR1pep peptide engagement for (A) CXCL8Synthetic or CXCL8MSynthetic in 50 mM phosphate, 150 mM NaCl, pH 6.5; (B) CXCL8 or CXCL8M in 50 mM phosphate, 150 mM NaCl, pH 6.5; (C) CXCL8 or CXCL8M in 50 mM HEPES, 50 mM NaCl, pH 7.0. All data shown within this figure were collected at 25 oC at 900 MHz.
30
Figure 7 CXCL8 engagement of hCXCR1pep leads to dynamic changes NMR amide nitrogen R2 relaxation rates (black lines) and R1 relaxation rates (green lines) were collected for the free CXCL8 (A) – left panel, free CXCL8M (A) – right panel, bound CXCL8 (B) – left panel and bound CXCL8M (B) – right panel. All data within this figure were collected in HEPES buffer at 25 oC and 600 MHz.
31
Figure 8 CXCL8 engagement of hCXCR1pep does not lead to dimer dissociation The extracted correlation times (τc) are plotted versus molecular weight (MW). The remaining values (black squares) are from previous studies that are also listed in Table 2. The MW values calculated from structures are listed in the table insert next to the MW values estimated from the experimental τc values. The MW, estimated from the experimental data was fit into a least-squares linear regression equation (τc = 0.59*MW).
32
Figure 9 CXCL8 is inherently flexible within the receptor binding site (A) Microsecond- millisecond (µs-ms) movements are located to the binding site as shown by the absolute value of H(S/M)QC exchange induced shifts. (B) Slow dynamics in wild type CXCL8 are primarily within the µs regime, as indicated by no R2-CPMG relaxation dispersion for representative residues (C50, E63, R26 and S30) (C) Structural summary of the detected motions within CXCL8, which include residues exhibiting elevated R1 relaxation rates that indicate local disorder on a fast timescale (green, see values in Fig. 3A) and H(S/M)QC exchange induced shifts that indicate exchange on the slow timescale (red). All residues mapped exhibited R1 rates and H(S/M)QC exchange induced shifts one standard deviation above the average. All NMR experiments within this figure were collected in HEPES buffer at 25 oC and 600 MHz.
33
Figure 10 CXCL8M weakly dimerizes
34
Figure 10 CXCL8M weakly dimerizes (A) R2-CPMG dispersion data for representative residues (C50, E63, R26 and S30) at two different protein concentrations (1 mM – blue, 0.5 mM – green) (B) The residues exhibiting elevated R2-CPMG dispersion values are mapped onto the structure. (C) R2-CPMG dispersion data for representative residues (C50, E63, R26 and S30) at two different fields (900 MHz – dashed green, 600 MHz – solid green). All NMR experiments within this figure were collected in HEPES buffer at 25 oC and at 600 MHz or 900 MHz.
35
Figure 11 Chemical shift-based calculations correspond to fast time scale dynamics and reveal that the C-terminal helix of the CXCL8 monomer partially unfolds
36
Figure 11 Chemical shift-based calculations correspond to fast time scale dynamics and reveal that the C-terminal helix of the CXCL8 monomer partially unfolds. (A) Secondary structure populations (SS%) calculated from the chemical shifts using the δ2D method (Camilloni et al. 37) for free CXCL8 (A – left panel) and CXCL8M (A – right panel); α-helix is shown in blue, β-sheet in red, polyproline II in green and random coil in black. (B) Local root mean square flactuations (RMSF, solid black lines) are calculated for each amide from the ensemble of structures using chemical shift-based distance restraints and are plotted for free CXCL8 (B – left panel) and free CXCL8M (B – right panel) in phosphate buffer. NMR amide relaxation rates R2 (dashed red lines) and R1 (solid red lines) collected for the free CXCL8 (B – left panel) free CXCL8M (B – right panel) in phosphate buffer. (C) Free energy landscapes as a function of the RMSD of the C-terminal α-helix (CT-Helix, x- axis) and the RMSD of the two loops corresponding to residues 28-31 and 50-53 (RMSD- Loops, y-axis) for free CXCL8 (C – left panel) and free CXCL8M (C – right panel). The RMSDs were calculated for each conformation in the ensembles generated here against structures available in PDB for the α-helix and two loops, respectively. The energy is in kJ/mol. (C – structure inserts) From the chemical shift-based calculations, sausage models were calculated based on the local RMSD for both free CXCL8 (C – structure inserts, left panel) and free CXCL8M (C – structure inserts, right panel). For clarity purposes residues 1- 7 are omitted from the MD structures. All data within this figure were collected in phosphate buffer at 25 oC at 600 MHz.
37
CHAPTER II
CD147 REGULATES CELLULAR METABOLISM VIA INTERACTION WITH SMALL
MOLECULE TRANSPORTERS IN PANCREATIC CANCER
Introduction and review of the literature
Pancreatic cancer
Pancreatic ductal adenocarcinoma (PDAC), the predominant form of pancreatic cancer, is the fourth leading cause of cancer related deaths in the United States with a 5- year survival rate of less than 6%57. Even with the recent development of new chemotherapies of combining gemcitabine with paclitaxel or the use of FOLFIRINOX, PDAC survival rates have not significantly increased within the last 30 years58, 59. Thus far the cause of PDAC has not been fully characterized although some recent evidence points to possible genetic predisposition leading to the frequent occurance of mutations in oncogenes and tumor suppressor genes58. Environmental risk factors have also been recognized including smoking, type II diabetes and excessive alcohol consumption58, 60. Chronic pancreatitis, which is described by long term inflammation of the pancreas is another risk factor suggested to be causative of pancreatic cancer61. Because of the lack of specific details on PDAC causes, the diagnosis of this cancer often does not occur until late stages of the disease, explaining the low survival rates.
PDAC typically starts in the ductal epithelium and progresses rapidly to other tissues including liver and spleen. One of the distinct features of PDAC is the appearance of desmoplastic stroma (highly fibrotic stroma). Indeed, one of the biggest challenges in targeting PDAC progression is the high level of stromal content in the tumor volume, with malignant cells comprising only 5-10% of the total tumor volume62. The major building block of stroma is the extracellular matrix (ECM), which is a dynamic three-dimensional structure composed of collagens, fibronectin, laminin, glycoproteins and proteoglycans. In addition to
38
the high ECM content, the stromal compartment is also composed of an assortment of diverse cell types including stellate, endothelial, nerve and immune cells. In order for cancer cells to metastasize to other tissues they have to disrupt the stromal compartment and, thus, stroma typically acts as a gatekeeper for the proliferative cells. Cancer metastasis requires cells to escape the primary tumor, survive blood stream entry and exit, and eventually develop into a tumor at a distal site. In PDAC, activation of oncogene pathways leads to increased invasiveness of tumor cells and a complex interplay between stromal microenvironment and cancer cells, called desmoplasia.
Increased cancer metastasis and invasion is often linked to the to epithelial to mesynchemal transition (EMT). EMT is one of the major hallmarks of cancer and it is a developmental process that allows normally non-motile cells to change their phenotype and increase motility (Figure 12). When cells undergo EMT they loose their epithelial features, which include cell polarity, cell-cell adhesion, non-invasion and chemo-sensitivity63. EMT leads to actin cytoskeleton rearrangement, increased cell-ECM interactions and enhanced migratory and invasive potential64. A reverse pathway, in which cells transition from a mesenchymal to epithelial (MET) phenotype can also take place and leads to a decline in proliferation and growth. One of the markers of EMT/MET is the E-cadherin protein, a calcium dependent cell-cell adhesion molecule downregulated or absent in mesenchymal cells. In epithelial cells, E-cadherin is mainly expressed on the membrane where it facilities cell adhesion via binding to the N-terminal region of another E-cadherin molecule expressed on neighborhood cells. The intracellular, C-terminal region of E-cadherin binds several other cytoplasmic proteins leading to translocation of those proteins to the cellular membrane.
This sequestration mechanism is especially important for β-catenin, a major component of the Wnt signaling pathway, typically upregulated in cancer. In mesenchymal cells β-catenin translocates to the nucleus to act as a transcription factor and activate transcription of proliferative genes. In contrast, membrane bound β-catenin is unable to translocate to the
39
nuclei and thus, is a marker of epithelial phenotype65. β-catenin localization can also be regulated by aberrant epidermal growth factor receptor (EGFR) signaling66. In oral and breast cancers, EGFR has been proposed to regulate β-catenin nuclear localization by preventing it from binding to E-cadherin67, 68. Another important marker of the EMT is
Vimentin, an intermediate filament protein ubiquitously expressed in mesenchymal cancer cells and downregulated when cells transition to epithelial phenotype. In addition to EGFR signaling, the expression of EMT/MET markers can be regulated by miRNAs, transcription factors and epigenetic marks69.
Using global genomic analysis, Jones et al. showed that the core signaling pathways altered in PDAC are Kristen rat sarcoma viral oncogene homolog (KRAS) signaling, apoptosis, cell cycle regulation, integrin signaling, adhesion, invasion, transforming growth factor β (TGF-β) signaling and Wnt/Notch pathway70. KRAS is a small GTPase important for the regulation of essential cellular processes especially cell growth and metabolic pathways.
It is well established that over 90% of PDACs exhibit activated KRAS mutations (G12D and
G12V), which drive pancreatic neoplasia; however, therapeutically targeting KRAS has proven widely unsuccessful71. A more feasible approach in inhibiting PDAC is to target
KRAS downstream signaling pathways. One such pathway is epidermal growth factor receptor (EGFR) signaling, which can induce cancer progression via an activation of EMT72-
74. Several other studies show that active KRAS drives metabolic reprograming in PDAC leading to increased aerobic glycolysis71, 75, 76. Targeting the specific metabolic changes associated with PDAC progression has recently become a very attractive therapeutic approach and although studies aided at characterizing the particular metabolic fluctuations are now emerging, more molecular details are required to better tailor those therapeutic strategies.
40
Cancer and cellular metabolism
One of the main hallmarks of cancer is the ability of tumor cells to increase their glucose uptake and escalate glycolysis to generate energy required for growth with an end point of lactate production from pyruvate, despite oxygen availability (Figure 13). This process, discovered by Otto Warburg and called the "Warburg effect", avoids the use of pyruvate for mitochondrial oxidative phosphorylation (OXPHOS) and subsequent energy production77. Although this is a highly inefficient pathway, with only 2 ATP produced per glucose molecule as opposed to 36 ATP produced in OXPHOS, it suppresses the oxidative stress response of mitochondria that could lead to cancer growth inhibition. Initially, increased aerobic glycolysis in cancer cells was associated with increased hypoxia (lack of oxygen) and acidity of tumor microenvironment. Hypoxia activates hypoxia responsive element α (HIF1α), which is a transcriptional factor responsible for regulating metabolic enzymes expression. Nevertheless, it is now well established that the initial increase in aerobic glycolysis persists in normoxic tumors78.
Because glucose-derived carbons are diverted away from OXPHOS, PDAC cells also rely on glutamine uptake to provide carbon for the maintenance of the TCA cycle. In normal cells, glucose derived-pyruvate is the primary source of carbon for fatty acid synthesis. In contrast, highly proliferating cells that undergo aerobic glycolysis utilize glucose for lactate synthesis and cells have to rely on other sources for lipid metabolism. In such an environment, glutamine becomes a carbon source for lipid generation via the mechanism of reductive carboxylation (Figure 14). Reductive carboxylation allows for the conversion of α-ketoglutarate (αKG) to citrate for subsequent lipogenesis, which is essentially a reverse step in TCA cycle. This mechanism is an alternative to oxidative carboxylation in which cells oxidatively metabolize glutamine derived- αKG through the forward steps of the TCA cycle (Figure 14)79. Recent studies suggest that reductive
41
carboxylation of glutamine is the predominant pathway utilized in highly proliferative cancer cells80.
The pentose phosphate pathway (PPP) is another vital biogenesis pathway, which takes place in the cytosol. PPP is important for producing NADPH and 5-carbon sugar precursors and is separated into two distinct branches, oxidative and non-oxidative.
Oxidative branch of PPP starts with conversion of glucose 6-phosphate via a two-step intermediate to ribulose 5-phosphate (Ru5P), which then feeds into the non-oxidative branch of PPP to produce other 5-carbon sugars. The non-oxidative branch of PPP is ubiquitously active and it uses intermediates of glycolysis; fructose-6phosphate (F6P) and glyceraldehyde-3phosphate (G3P). The oxidative PPP plays very important role in maintaining cellular redox balance through the generation of NADPH81. The rate-limiting factor in regulating oxidative PPP is the expression and activity of glucose 6-phosphate dehydrogenase (G6PD), which catalyzes the first reaction in this pathway. Studies show that oxidative PPP is often upregulated in highly proliferating cells (i.e., cancer cells) where the
NADPH produced in this pathway is used for detoxification of reductive oxygen species
(ROS)82.
CD147
Extracellular matrix metalloproteinase inducer (CD147, also known as EMMPRIN) is a highly glycosylated type I single pass transmembrane protein upregulated in a variety of cancers including pancreatic cancer83, 84. CD147 was initially discovered by several independent groups and was given a variety of names including gp42, HT7, OX-47, M684, 85.
In addition, its gene and sometimes protein are called basigin (BSG)86. CD147 has been implicated in the regulation of leukocyte differentiation hence, its most commonly used name, CD147, comes from Cluster of Differentiation Nomenclature. CD147 is widely expressed amongst all vertebrates and in humans its gene is located on chromosome 19 at
42
p13.387. The encoded protein is a member of the immunoglobulin (Ig) superfamily, where it forms a distinct family of immunoglobins, along with embigen and neroplastin, within the very broad Ig superfamily88. CD147 was initially characterized as a matrix metalloproteinase
(MMP) inducer, although several other functions have been discovered by now, including regulation of cell proliferation, neuronal network development, spermatogenesis, metastasis, differentiation, β-amyloid production and tumor cell migration84, 86, 89. While numerous roles for CD147 in cancer cell malignancy have been demonstrated, minimal work has been published regarding the effects of CD147 on cellular metabolism.
There are two different CD147 isoforms present in humans, a retina specific isoform-
1 and the highly ubiquitous isform-2 expressed in almost all other tissues. Both isoforms are composed of a 23 residue cytoplasmic tail (CT), 41 aa transmembrane (TM) region and an extracellular region (ectodomain or ECD), which differs between the two isoforms (Figure
15A). Isoform-1 contains three Ig like domains (Ig0, Ig1 and Ig2), while isoform-2 (herein referred as CD147) is smaller in size due to the absence of Ig0. CD147 is composed of 269 amino acids (28 kDa) with additional 30-40 kDa of glycosylation present. Two different glycosylated forms of CD147 have been identified; lowly glycosylated (LG) and highly glycosylated (HG), with the latter one being a mature fully active form of CD14790, 91. All sites of glycosylation are contained within the extracellular region, one on Ig1 (N44) and two on Ig2 (N152 and N186). The Ig0 domain has been proposed to contain glycosylation sites but thus far there is no comprehensive analysis of such post-translational modification86, 92.
Little is known about Ig1-Ig2 glycosylation in general. In particular, Tang et al. showed polylactosamine-type sugars to comprise CD147 glycosylation, while other studies also proposed the presence of complex carbohydrate chains containing sialic acid, galactose and
N-acetylglucoseamine93, 94. Recent literature suggests that the difference in glycosylation pattern might be delimited by CD147’s direct or indirect interactions with caveolin, integrins and Galectin-395 96. This association of CD147 with other proteins is thought to prevent the
43
formation of highly glycosylated CD147 and therefore influence its activity93. The proper glycosylation pattern might play a critical role for its membrane translocation mechanism as well. Huang et al. revealed that SMMC7721 (human liver hepatocarcinoma) cells transfected with a mutated form of CD147 incapable of glycosylation (N44Q, N152Q and N186Q) were unable to translocate CD147 to the membrane94. This observation suggests that the glycosylation mechanism, which takes place in the ER and Golgi, is critical for the subsequent cellular translocation of CD147 to the plasma membrane.
CD147 can be released by cells to the extracellular region via microvesicle shedding
(full length protein) or cleavage of the extracellular region (ectodomain). Microvesicle shedding allows for secretion of full length membrane embedded protein that can then have stimulatory activities on other cells97. A second mechanism of CD147 release is via cleavage of the ectodomain from the membrane by MMPs84, 98. CD147 ectodomain (referred within as
CD147-ECD) stimulates the expression of MMPs and pro-inflammatory cytokines22, 99.
Multiple lines of evidence suggest that extracellular forms of glycosylated CD147 stimulate their own expression and consequently high levels of both MMPs and cytokines are continually secreted due to the release of extracellular CD14798, 100-102. MMPs released via
CD147 stimulation can then cleave CD147 from membranes thereby forming a positive feedback loop98. Un-glycosylated form of CD147, purified from E.coli also has stimulatory activities but the extent of such functions is significantly reduced presumably due to the lack of glycosylation92. There is some evidence in the literature suggesting that membrane bound
CD147 can act as a receptor for its extracellular form53. However, this would likely require a homophylic interaction between the Ig domains, an event disputed by our group using the unglycosylated CD14799, 103. Though it is possible that such interaction is mediated by the presence of glycans, further studies are needed to clarify such theory.
Studies performed in CD147 knockout animals (CD147-/-) have shed light on the importance of CD147 expression in normal tissues. For example, CD147 deficient animals
44
were defective in the regulation of spermatogenesis, lymphocyte responsiveness and neuronal functions104. The survival of CD147-/- mice was low with all surviving animals being infertile105. CD147 also plays an important role in adhesion of neighboring cells. For example in U937 cells, CD147 has been proposed to regulate cell aggregation via the leukocyte function antigen-1 (LFA-1) and intracellular adhesion molecule-1 (ICAM-1) pathway106.
Although several different functions for CD147 have been proposed, its role in promoting tumorigenesis has been given the most attention. This is in agreement with high levels of
CD147 detected in most cancers. In lung cancer, CD147 overexpression leads to acceleration of cancer cell growth, proliferation, migration and anchorage-independent growth in a Wnt/β-catenin dependent matter107. In lymphoma and breast cancer cells, the localization and homodimerization of multidrug resistance (MDR) protein is regulated by
CD147, suggesting a contribution of CD147 to the regulation of chemosensitivity108, 109.
Furthermore, multiple lines of evidence uncovered that CD147 silencing in pancreatic cancer leads to inhibition of malignant potential and cell invasion but the mechanism of this activity has not been fully explained110, 111.
CD147 interacting partners
Numerous proteins have been proposed to interact with CD147, but only a few of these interactions have been studied at the molecular detail (Figure 15B). For example, a variety of integrins (integrin α3β1, integrin α6β1 and integrin β1) are implicated in an interaction with CD147, but it is currently unclear if those are direct interactions or associations mediated by other proteins95. Several other proteins (described below) have been shown to interact with integrins, and, thus the possibility of a large complex formation between CD147 and its proposed targets is likely to be true. CD44, a cell surface glycoprotein and hyaluronan receptor, was previously detected in a large complex with
CD147 and EGFR in lipid-rafts. CD147 interaction with CD44 is also attributed to CD147's
45
ability to regulate hyaluronan levels in breast cancer cells112-114. The progression of
Alzheimer’s disease has been linked to CD147 interaction with γ-secretase, which cleaves the β-amyloid protein leading to the accumulation of β-amyloid peptides115. A possible interaction of CD147 with γ-catenin, an endothelial junction protein and nucleotide-binding oligomerization domain-containing protein 2 (NOD2), an intracellular receptor for listeria infection, has also been noted116, 117.
Multiple extracellular proteins are proposed to utilize CD147 as their receptors.
Extracellular Cyclophilin A (CypA, extensively reviewed in chapter III) is probably the most studied CD147 ligand. Our group and others performed biophysical and biochemical studies showing that CypA can bind, with a relatively weak affinity to the extracellular region of
CD14799, 118, 119. In addition to CypA, another extracellular cyclophilin, Cyclophilin B (CypB) can bind to CD14785. In each case, the interaction with CD147 has been proposed to play a critical role in the functional activities of those proteins especially in regard to their chemotactic, viral infection and other functions85, 120. Depending on the cellular system studied, platelet glycoprotein VI (GPVI), S100A9 and apolipoprotein D have all been shown to signal in a CD147 dependent manner121-123. Using chemical cross-linking and pull-downs combined with LC-MS/MS, Xu et al. revealed that CD147 interacts with the ectodomain of
4F2 heavy chain (CD98hc), however, there are no biophysical studies confirming a direct interaction124. CD98hc is a heterodimeric transmembrane protein, which forms a complex with large neutral amino acid transporter 1 (LAT1) and also binds another possible CD147 interacting partner, integrin β1. In addition, the same group showed evidence of CD147 interaction with neutral amino acid transporter (ASCT2), LAT1, epithelial cell adhesion molecule (EpCAM) and monocarboxylate transporter I (MCT1). Although, several CD147 interacting partners have been identified, the data regarding the specific molecular details of those interactions remains unclear.
46
Another class of CD147 interacting proteins are monocarboxylate transporters
(MCTs); MCT1 and MCT4, specifically125. MCTs are multipass transmembrane proteins responsible for transport of small molecules (e.g. lactate, pyruvate, hormones etc.) across the plasma membrane and, thus, the regulation of cellular metabolic processes126. MCTs belong to the superfamily of solute carriers (SLCs), which is composed of almost 300 members and is divided into 52 families127. SLCs are multi-pass transmembrane proteins localized to the plasma membrane with only a few members located on other cellular members. MCTs form the SLC16 family with 14 members, which all comprise 12-pass transmembrane proteins with conserved sequence motifs. MCT1 and MCT4 are expressed in several tissues, with MCT4 a hypoxia-inducible protein overexpressed in cells particularly reliant on glycolytic processes, especially malignant tumor cells126. PDAC cells, due to their high metabolic dependence, tend to rely more on MCT4 expression for lactate export as opposed to MCT1. MCT1 is ubiquitously expressed in all tissues and is responsible for shuttling lactate in and out of the cells depending on the cellular requirements. Recent discoveries indicate that the expression of MCT1 and MCT4 is dependent on CD147, although the role of CD147/MCT interaction in regard to PDAC progression has not been fully depicted125, 128. Kirk et al. proposed that CD147 might be a chaperone for MCT1 and
MCT4, since both of those MCTs were unable to translocate to the plasma membrane in the absence of CD147125. Furthermore, the same study provided evidence that the CD147 ectodomain does not play a role in such chaperone activity, suggesting that the CD147 transmembrane and/or C-terminal regions mediate its interaction with MCTs. An elegant study by Baek et al. showed that silencing of MCT4 in pancreatic cancer cells had only modest influence on CD147 expression. Although depletion of MCT4 had a striking effect on cell viability, CD147 depletion in those same cells did not recapitulate that phenotype129.
Those observations suggest that although CD147 can function cooperatively with MCTs,
47
there might be other factors that are important for CD147 function that are independent of its interaction with MCTs.
Study rationale
An emerging number of studies have indicated that the progression of PDAC is heavily reliant on the regulation of the EMT and metabolic pathways. In addition, CD147 overexpression promotes cancerous phenotypes, however the specific mechanistic details related to the pathways under CD147 regulation remain elusive. Likewise, the existence of different signaling pathways regulated by the full-length protein or the extracellular region alone requires further characterization. The described here study aims to decouple the specific signaling pathways regulated by the different forms of CD147, particularly since
CD147 exists in both forms in the biological systems. Another important aspect of CD147 signaling is the presence of high level of glycosylation in cancerous cells. The contribution of
CD147 glycans has been linked to its cellular membrane trans-location or ectodomain activity though, there is an overall lack of glycan contribution analysis in regard to PDAC progression. Lastly, the molecular details of CD147 interactions with other proteins and functional consequences of such interactions have not been fully characterized. Thus, the focus of this study is to characterize the mechanism of CD147 activity in pancreatic cancer, establish the role of glycans in its signaling, and unravel the specific signaling pathways regulated by CD147.
Results
The extracellular region of CD147 exhibits little to no activity
The extracellular form of CD147 (amino acids 22-205) has been proposed to retain a biological activity, in regard to inducing MMPs secretion, after being cleaved from the membrane bound CD14794. To test this possibility, we purified CD147-ECD from E. coli and used it to treat different pancreatic cancer cell lines. We detected stimulatory activity in
48
regard to cytokine and MMP9 secretion, but only when very high (above 25 µM) concentrations of CD147-ECD were used (Figure 16). The need for high concentration of
CD147 could be explained by the requirement of the glycosylation for its activity. Indeed, a biological form of CD147 is a highly glycosylated protein with three sites of glycosylation, all within the extracellular region and, hence, the activity of this protein should be measured in regard to its biologically relevant form i.e., glycosylated CD147-ECD (CD147-ECDgly). Thus, to generate the glycosylated CD147 ectodomain we utilized a mammalian expression system for high yield protein expression (Figure 17). We used mammalian human embryotic kidney cells (HEK293) optimized for high-density suspension growth in combination with transient transfection and cell cycle inhibition for increased mRNA production130-134. This system allowed us to purify the glycosylated form of CD147 containing both Ig domains that house all three sites of glycosylation (Figure 18). We also purified the unglycosylated form of
CD147, from the same expression system by mutating all three glycosylation sites to glutamine residues (N44Q, N156Q and N186Q) in order to prevent endogenous addition of
N-linked glycans.
We then tested the activity of CD147-ECDgly in regard to MMPs and cytokine secretion in pancreatic cancer cell lines by performing multiple broad-based screens with varying concentrations of the purified protein and different treatment times. Using ELISA based methods, an array (10-plex pro-inflammatory human array, Meso Scale Discovery) of pro-inflammatory cytokines (IFNγ, IL10, IL12 p70, IL13, IL2, IL4, IL5, IL8, TNFα and IL1b) and MMPs (MMP9, MMP3 and MMP2) were tested. In addition, we performed qRT-PCR screens and MMP2 and MMP9 activity assays. In each case, only very modest increases or no stimulatory activity for CD147-ECDgly was observed (representative data shown in Figure
19). To exclude the possibility that CD147-ECDgly might not be active in regard to pancreatic cancer cells, but might instead exhibit activity in other systems, we also tested other cell lines, including monocytic cells (U937), embryotic kidney cells (HEK293), colon cancer cells
49
(Caco-2) and cervical cancer cells (HeLa). Consistent with the results in pancreatic cancer cell lines we did not observe any stimulatory activity for CD147-ECDgly in other cellular systems (Figure 19). These results are also in agreement with published studies where only limited data is available showing activity for CD147-ECDgly.
CD147 downregulation in pancreatic cancer cells inhibits cancerous phenotype
Based on the previous results that indicated little to no stimulatory activity for the glycosylated CD147 ectodomain alone, we postulated that it is the full-length protein that is functionally important in regard to PDAC progression. To characterize the function of CD147 in PDAC, we evaluated the expression of CD147 in several PDAC cell lines and three immortalized non-cancerous pancreatic cell lines. CD147 is highly expressed in most cells including non-cancerous pancreatic cells (Figure 20). In addition, cancerous cell lines exhibit
CD147 signal at higher molecular weights (35-65 kDa) compared to the non-cancerous cells
(35-45 kDa), indicating increased level of glycosylation in those cells. Considering that glycosylation is the only known post-translational modification that occurs for CD147, this indicates that there is an increased level of glycosylation of CD147 in cancer. This observation is consistent with an increase in the glycosylation pathway commonly associated with more cancerous phenotypes135.
Because aggressive cancer is often accompanied by increase in mesenchymal phenotype and CD147 has been proposed to regulate the EMT, we proceeded to test the expression of EMT markers (E-cadherin, β-catenin and Vimentin) in those same cell lines
(Figure 20). The different expression of EMT marker proteins suggests different cell morphology and behavior, which can classify cells into two different categories, mesenchymal or epithelial cells. Due to previously established diverse roles for CD147 in the regulation of cancer progression and the need to obtain more details into the specifics of
CD147 activities, we selected cell lines characterized by different EMT phenotype for further
50
analysis. Specifically, we used two mesenchymal cell lines, PANC1 and MiaPaCa2 and one epithelial cell line, L3.6pl (all carrying activating KRAS mutations)136, 137. Initially we attempted to generate a CD147 knock-out cells using CRISPR/Cas9 system, however, we were not able to isolate viable cells harboring complete depletion of CD147 (data not shown). Although, such attempts have previously been reported in some cancer systems the lack of success in our case could potentially be explained by the importance of CD147 in
PDAC progression128. However, with the use of three shRNAs, each targeting different regions of CD147 gene, we generated stable knockdown cells with varying levels of CD147 downregulation (Figure 21A). The level of CD147 downregulation was consistent with concomitant inhibition of cell growth and revealed the most significant cell growth suppression in cells with the highest CD147 depletion (Figure 21B).
We next screened the CD147 depleted cells for changes in the expression of EMT marker proteins, since aberrant EMT regulation is often associated with decrease in cell growth and invasion. We observed an inhibition of the EMT in cells exhibiting a mesenchymal phenotype, especially evident in PANC1 cells and to a lesser extent in
MiaPaCa2 (Figure 22). As expected, cells exhibiting an epithelial phenotype (L3.6pl) showed no significant increase in epithelial markers (Figure 22). PANC1 cells also exhibited an increased expression of membrane bound β-catenin (epithelial marker) as illustrated by
IF (Figure 22B). In addition, we tested the migratory capabilities of CD147 depleted cells, since decrease in migration is associated with reversal of the EMT phenotype. Both PANC1 and MiaPaCa2 cells that were depleted of CD147 demonstrated reduced migration rates in response to serum stimulation with a very modest decrease in L3.6pl CD147 depleted cells, which do not exhibit high level of migration due to their epithelial phenotype (Figure 23).
Those observations suggest that CD147 expression plays a critical role in regulating PDAC progression and is consistent with other studies that illustrated the involvement of CD147 in regulating cancerous phenotypes83, 84, 128.
51
SILAC analysis reveals differential expression of several metabolic and adhesion proteins in CD147 depleted cells
The striking influence of CD147 downregulation on cell growth and migration led us to investigate the global proteome changes upon CD147 depletion in PDAC cells, via stable isotope labeling by amino acid in cell culture (SILAC). In order to better represent the overall role of CD147 in PDAC regulation we used cells exhibiting the most significant CD147 depletion (shCD147 #1, here referred as shCD147) from two cell types, PANC1 and L3.6pl, which show very different phenotypic changes due to CD147 depletion. We cultured CD147 knockdown cells in heavy labeled media (13C-lysine and 13C-arginine) and control cells in light media (12C-lysine and 12C-arginine) and analyzed global protein expression changes using mass spectrometry (MS). The ratio of heavy/light (shCD147/shCTRL) was used for fold changes quantification and was normalized to total peptide count. We identified a total of 87 and 51 proteins differentially regulated in PANC1 and L3.6pl cells, respectively (Figure
24A). Out of this group, a significantly higher number of proteins were downregulated, including CD147 and its proposed interacting partner, MCT4. Consistent with the extent of phenotypic changes observed in L3.6pl cells, these cells had a lesser number of differentially expressed proteins with 7 proteins present in both PANC1 and L3.6pl cells
(Figure 24B, marked with a star).
To identify the specific pathways under CD147 regulation we performed functional analysis of all statistically deregulated proteins identified in SILAC experiments. Gene annotation (GO) analysis revealed that the vast majority of the analyzed proteins are involved in several pathways implicated in the regulation of adhesion, cell motion and metabolic processes (Figure 24C). Since modification of cell motion and adhesion are both hallmarks of the EMT, the deregulation of proteins involved in those processes is consistent with the observed phenotypic changes in shCD147 cells, particularly PANC1 cells. For
52
example, transglutaminase 2 (TGM2), integrin α2 (ITGA2), aldehyde dehydrogenase 1 A3
(ALDH1A3) and collagen α1 (COL1A1) were all downregulated in CD147 depleted cells.
This is in agreement with published studies showing a direct correlation between the expression of the indicated proteins and downregulation of E-cadherin and mesenchymal characteristics138-141. In addition, several actin binding and remodeling proteins were upregulated in CD147 depleted cells, which Includes, Ca regulated gelsolin (GSN) and non- muscle heavy chain myosin II B (myosin II B, gene name: MYH10), typically downregulated in pancreatic cancer and highly expressed in normal tissues142, 143. In addition Annexin 1, also upregulated upon CD147 depletion, has been recently proposed to negatively regulate the EMT via its interaction with actin filaments144, 145. Moreover integrin β4 (gene name
ITGB4) expression was highly augmented in CD147 depleted cells. Integrin β4 is a receptor for laminins and a critical mediator of cell-cell and cell-ECM contacts.
Furthermore, we observed several proteins involved in the regulation of metabolic processes to be differentially expressed in CD147 knockdown cells. MCT4 (gene name:
SLC16A3), the proposed CD147 interacting partner responsible for cellular lactate export, was one of the identified proteins in this group. Several enzymes involved in the regulation of cellular metabolic pathways were deregulated upon CD147 depletion. In particular, pyruvate kinase isoenzyme 2 (PKM2) and glucose-6-phosphate dehydrogenase (G6PD) important for regulating the conversion of PEP to pyruvate and the initial step of oxidative
PPP, respectively, were both downregulated with CD147 depletion. Several proteins critical for the activation of glycolysis were upregulated including, glucose transporter 1 (GLUT1, gene name SLC2A1) and hexokinase containing domain 1 (HKDC1). Furthermore, enzymes implicated in serine biosynthesis (phosphoglycerate dehydrogenase, PHGDH) and arginine biosynthesis, and proline biosynthesis (ornithine aminotransferase, OAT and pyrroline-5- carboxylate reductase 1, PYCR1) also exhibited increased expression.
53
MCT1 and MCT4 expression contributes to CD147 activity but does not fully explain
CD147 mediated PDAC phenotypes
Based on our global proteome data monitored through SILAC experiments, two major processes appeared to be highly deregulated upon the loss of CD147 expression; cellular adhesion and cellular metabolism. Both adhesion and metabolism are implicated in the regulation of cell growth and the EMT, phenotypes affected by CD147 depletion. In addition, several lines of evidence suggest an interaction between CD147 and two members of the MCT family, MCT1 and MCT4, which are both involved in lactate transport. In order to identify the specific mechanism of CD147 function in PDAC, we proceeded to test the contribution of MCT1 and MCT4 interactions with CD147 to the identified phenotypes. As illustrated in Figure 25, both MCTs were downregulated in CD147 depleted cells, with the highest level of downregulation in PANC1 shCD147 #1 and #3, which also exhibits the highest level of CD147 depletion (Figure 21). Noteworthy, the decreased expression of
MCT4 was also identified in SILAC experiments; however, we did not detect MCT1 in the same experiments. This is most likely due to a much higher expression of MCT4 in PDAC cells, as compared to MCT1. Although, MCT1 is ubiquitously expressed, MCT4 it typically overexpressed in cancer tissues and its upregulation is correlated with apparent hypoxic and highly glycolytic tumors, which is generally found in PDAC tumors128, 129, 146.
To further characterize the relationship between CD147 and MCTs, we generated
MCT4 depletion cell lines in two representative PDAC cell types (PANC1 and L3.6pl). Cells harboring MCT4 depletion displayed significant decrease in CD147 expression with no effect on MCT1 expression, but substantial decline in cell growth (Figure 26A and C). Next, we tested the contribution of MCT4 downregulation to EMT mediated phenotypic changes via WBs. As shown in Figure 26B, there was no observable variation in EMT markers expression upon MCT4 depletion, suggesting that MCT4 is not involved in the EMT
54
regulation, as is CD147. This observation is consistent with previously published results where MCT4 knockdown led to inhibition of cell growth without any impact on EMT in PDAC cells129.
MCT4 is a proton coupled lactate transporter, and hence, decreased MCT4 expression should be accompanied by intracellular lactate accumulation. To test this presumption, we next investigated the functional consequences of MCT4 depletion by measuring cellular lactate levels. For this analysis we utilized three different sets of conditions: (i) CD147 depleted cells, (ii) MCT4 depleted cells, and (iii) MCT4 depleted cells treated with MCT1 inhibitor (ARC-C155858, referred in here as MCT1inh). The treatment of shMCT4 cells with MCT1inh allowed us to account for inhibition of both MCTs activity of lactate efflux, which would be the case in shCD147 cells where both MCTs are downregulated. MCT1inh specifically inhibits MCT1 and MCT2 without affecting MCT4 activity147. MCT2 is a high affinity pyruvate transporter expressed at low levels in pancreas148. We performed MS based metabolomics analysis of cellular extracts from the indicated cells to detect changes in intracellular lactate levels. We found significant accumulation of intracellular lactate in shCD147 PANC1 cells with only a modest increase in shCD147 L3.6pl cells (Figure 27); which is in agreement with the extent of CD147 mediated phenotypic changes observed in these cells. In addition, MCT4 depletion did not lead to a significant accumulation of intracellular lactate possibly due to compensatory activity of
MCT1, which was inhibited by treatment of MCT4 knockdown cells with MCT1inh.
The use of high-resolution hybrid quadrupole orbitrap mass spectrometer coupled with ultra high performance liquid chromatography (UHPLC) for global metabolomic analysis allowed us to characterize relative levels of several metabolites in the above mentioned samples. We compared the relative abundance levels of a subset of metabolites, in shCD147, shMCT4 and shMCT4 cells treated with MCT1inh and, as it can be depicted in
Figure 28, changes in the identified metabolites do not follow the same trend between
55
shCD147 and the different shMCT4 cells. This observation suggests that CD147 mediated signaling depends not only on MCT4 or MCT1 expression and activity, but possibly other proteins or signaling pathways are involved. This is also in agreement with the results shown above where different phenotypic changes were detected in MCT4 depleted cells as opposed to cells harboring CD147 knockdown.
MS based cross-linking experiments identify adhesion and transport proteins as possible CD147 interacting partners
Because MCT1 and MCT4 depletion and inhibition, respectively, do not recapitulate the effects of CD147 depletion in PDAC progression, we pursued the possibility of other
CD147 interactions being involved in CD147 mediated signaling. We used a cross- linking/pull-down method followed by MS to identify the potential CD147 interacting partners.
We chose PANC1 cells, since these cells exhibited the most significant phenotypic changes due to CD147 depletion. Cells were transfected with wild type CD147-GFP (C-terminal fusion) or GFP alone and cross-linked with the cell permeable cross-linking reagent, disuccinimidyl suberate (DSS). DSS is a homobifunctional cross-liker, which reacts with primary amines, typically lysines and thus allows for cross-linking of lysines that are no further than 11.4 Å (spacer arm length). Following cross-linking cells were lysed and proteins pull-downed with GFP nanobody conjugated to sepharose beads for subsequent
MS analysis of eluted proteins (Figure 29).
As illustrated in Figure 30A, our cross-linking/MS method allowed for straightforward identification of CD147 interactions (CD147 WB, GFP lines compared to CD147-GFP lines).
The MS analysis identified a total of 1469 proteins, whereby 747 proteins were present in both GFP and CD147-GFP fractions with 153 proteins specific for the CD147-GFP pull- downs. To select for the possible interaction partners and exclude false positives, only proteins that were present in both biological replicates and were detected in CD147-GFP ±
56
DSS with at least three unique peptides were chosen for further analysis (Table 4).
Furthermore, we only observed 16 proteins in the non-cross-linked samples, as opposed to
153 in cross-linked samples. This observation could be due to the appearance of weak or transient interactions between CD147 and its partners, or it could be due to experimental issues i.e. peptide mass modification due to crosslinking, poorly cleavable peptides.
A small number of proteins detected in the cross-linking/pull-down experiments have previously been reported (Table 4 marked with a hash tag). Xu et al. published an MS study with chemical cross-linking using three different cell lines but only identified 19 possible interactions124. However, both this prior study and our analysis detected three members of the MCT family present in CD147 pull-downs. Specifically, MCT1, MCT4 and MCT8 were all amongst pull-downed proteins with MCT4 detected in non-cross-linked samples as well.
MCT8 is a transporter primarily expressed in the thyroid, but its expression is low in PDAC cells, consistent with the low number of peptides identified for this protein. Again, we detected less peptides for MCT1 than for MCT4, supporting the supposition that MCT1 is expressed at lower levels in PANC1 cells.
To further analyze the possible CD147 interactions, we performed a functional analysis of all 153 potential proteins detected in cross-linking/MS analysis to identify any common functional trends. GO term analysis revealed that the majority of the classified proteins are involved in cellular transport processes (Figure 30B), which is also consistent with our SILAC data that revealed alterations in proteins integrally involved in cellular metabolism. Specifically, several of the proteins (23) belong to the SLC family, which also includes MCTs. In general, SLCs are integral membrane proteins important for transport of variety of small molecules, including; metabolites, amino acids, hormones and other solutes.
In addition, we identified a significant number of proteins (18) implicated in the regulation of cell adhesion. Interestingly, deregulation of cell adhesion was also detected in the functional
57
analysis of SILAC proteins, suggesting a link between CD147 interactions and CD147 signaling.
Cross-linking/MS data can also be used to identify the particular interaction sites or regions of close proximity. The correct identification of cross-linked peptides can be quite challenging in large MS data sets and, therefore, the interpretation of such searches must be further validated. However, even after further validation caution should be taken since each cross-linked product contains two covalently attached peptides with uneven fragmentation pattern. Taking this into account, data obtained from cross-linked searches can still be valuable especially as an intermediate step before further validation. We searched our MS data set for the presence of any cross-linked peptides between CD147 and any of the 153 proteins. Peak lists from the individual MS runs were created in
MASCOT and analyzed using MS-Batch software available from Protein Prospector platform. We performed searches with a full list of 153 proteins identified in MS cross-linking analysis plus CD147 itself. The identified cross-linked peptides were scored using false discovery rate (FDR). FDR generated in those searches uses decoy database strategies to reduce the amount of false positives149.
Our cross-linking searches led to the identification of 8 possible interactions (Table
5). Interestingly, we did not detect any cross-linked peptides between CD147 and any of the
MCTs. This lack of cross-linked peptides could be due to the nature of CD147 interaction with MCTs, which has been proposed to be through the transmembrane region or cytoplasmic tail125. CD147, similarly to numerous other transmembrane proteins does not contain any lysines within the membrane-spanning domain and therefore our method is incapable of detecting interactions taking place in this region. It is also important to point out that none of the identified cross-linked peptides were detected in more than one experiment, although we repeated the study with a second set of independent biological replicates. The inability to replicate our identification of cross-linked peptides most likely comes from the
58
complexity of the analyzed samples. Interestingly, we identified cross-links between CD147 and plasma membrane Ca-transporting ATPase 1 (PMCA1, gene name: AT2B1). PMCA1, a
135 kDa integral transmembrane ATPase responsible for Ca efflux to the extracellular region, was the highest scoring cross-linked peptide identified. Recent studies show that
PMCA isoforms are critical for Ca homeostasis in PDAC and their activity depends on ATP availability from glycolytic processes upregulated in cancer cells150, 151. In addition, PMCA1 has also been detected in a parallel MS cross-linking study and a different isoform of PMCA,
PMCA4 was recently identified to interact with CD147 in T cells124, 152.
The conformation of possible CD147 interacting partners
The presence of a protein in a pull down samples does not unambiguously mean that an interaction is taking place. Even with the use of stringent selection criteria, it is possible for some proteins to be identified due to non-specific interactions, large complex presence, or redundancy of the peptides in MS analysis. To further confirm some of the possible interactions we focused on proteins that were identified with high likelihood of association with CD147 based on several criteria. These criteria include: enrichment in CD147-GFP
±DSS samples, presence of a cross-linked peptide, previous evidence of such interactions in the literature, or the downregulation of expression identified in SILAC experiments. We selected MCT1, MCT4, PMCA1, ASCT2 (gene name: SLC1A5), LAT1 (gene name
SLC7A5), integrin β1 (ITGB1), multidrug resistance protein 1 (MRP1, gene name: ABCC1), intracellular adhesion molecule 1 (ICAM1 or CD54, gene name: ICAM1), CD98hc and
EGFR.
We first confirmed the specificity of the interactions by WBs with pull-down samples from cells transfected with CD147-GFP or GFP (control) and treated with ± DSS (DMSO was used as a no-treatment control). We did not observe any signal in the cross-linked samples (input or pull-down) for CD98hc (Figure 31A) implying that the antibody epitope
59
was lost due to cross-linking and, thus, we could not conclude if this interaction takes place in our cells. We did, however, test the interaction of CD98hc ectodomain with CD147-ECD since such association has already been suggested124. We expressed and purified isotopically labeled CD147-ECD and unlabeled CD98hc-ECD and used NMR titrations to test for an interaction. Using chemical shift perturbations analysis (Figure 31B), we observed no interaction between CD147-ECD and the extracellular region of CD98hc. This is inconsistent with previously reported interaction though, the data reported by Xu et al., did not indicate a direct engagement, as no biophysical analysis was used to test for such interaction124. It is also possible that the CD98hc interaction with CD147 is through the TM or C-terminal region or there are other proteins that mediate this association e.g. LAT1,
CD98hc interacting partner.
WB analysis did not detect signal for ITGB1, MRP and EGFR, but we were able to confirm interactions for other proteins tested (Figure 32). Specifically, with the exception of
LAT1, we observed bands for all proteins (MCT4, MCT1, ASCT2, PMCA1 and ICAM1) at the expected molecular weights for the cross-linked complexes of CD147-GFP. This is also consistent with the molecular weight of CD147 observed in CD147-GFP cross-linked fractions detected in CD147 WBs (Figure 30), where there is a significant amount of signal observed at higher molecular weight. The lack of LAT1 cross-linked complex is most likely due to the previously characterized interaction of LAT1 with CD98hc, rather than direct interaction with CD147 (Figure 32F)153. We also observed signal for MCT1 and MCT4 at lower molecular weights (Figure 32A and B), which is most likely due to the multimeric MCT interactions reported previously for both of those proteins154.
We further tested the interactions in an endogenous system without the need to overexpress any of the proteins. We used the in situ proximity ligation assay (PLA), which expands on the traditional immunofluorescence to allow for detection of protein-protein interactions with high sensitivity and specificity. PLA has previously been used to
60
characterize an interaction between EGFR and CD147 in breast cancer cell lines, hence we also utilized this assay to test for the detected inetarctions112. As illustrated in Figure 33A, we confirmed interactions between CD147 and a subset of proteins and as expected, these interactions were lost or diminished in CD147 depleted cells, as it would be expected. We next sought to examine the expression levels of those confirmed CD147 interacting partners upon CD147 depletion. Interestingly, several of the identified proteins were highly downregulated due to diminution of CD147 expression while other proteins that we tested were upregulated. In particular, MCT4, MCT1, PMCA1 and ICAM1 were all downregulated upon CD147 depletion as shown in Figure 33B and C. In contrast, ASCT2, LAT1 and
CD98hc were upregulated upon CD147 depletion (Figure 33B). Although, surprising at first this observation might be due to the global phenotypic changes associated with CD147 signaling. It is possible that CD147 interactions with the downregulated proteins is required for their expression while the upregulated proteins require other secondary factors that are not interacting directly with CD147. Those other proteins might be upregulated in shCD147 cells due to the observed phenotypic changes and, thus, CD147 might be negatively influencing the expression of some of the possible interacting partners.
CD147 acts as a chaperone for several transmembrane proteins
Our findings illustrate that CD147 interacts with a number of membrane proteins and the expression and membrane localization of those proteins is significantly suppressed upon
CD147 deletion (Figure 33C). This discovery led us to further investigate the specifics of those interactions and their functional consequences. CD147 was previously shown to be important for accompanying MCTs (MCT1 and MCT4) to the membrane and, thereby, assure their proper cellular localization. This role of CD147 as an ancillary protein for MCT1 and MCT4 expression and cellular membrane localization could be explained by CD147's ability to be a chaperone for those proteins. A chaperone is any protein that assists another
61
protein in folding, translocation or protects it from degradation155. Thus, to determine whether CD147 plays a chaperone function for its interacting partners, or if it is simply an ancillary protein we tested some of the proposed chaperone functions e.g. translocation and protection from degradation. If CD147 is indeed a chaperone for its interacting partners, one would predict that in the absence of CD147 those proteins would undergo proteasomal degradation. To test this hypothesis we treated shCTRL or shCD147 cells with increasing concentrations of a proteasomal inhibitor (MG132) and measured the changes in its interacting protein expression by WBs. Figure 34A illustrates that we were able to restore
MCT4, MCT1, PMCA1 and ICAM1 expression when proteasomal degradation is inhibited, suggesting that CD147 protects its interacting partners from degradation. Furthermore, we demonstrated by immunofluorescence that although inhibition of proteasomal degradation restores CD147 interacting partners expression, these proteins are unable to translocate to the cellular membrane in the absence of CD147 (Figure 34B). This inability of the indicated proteins to translocate to the membrane, in the absence of CD147, implies that CD147 is necessary for their translocation event. Collectively, these data show that CD147 protects its targets from degradation and translocates them to the cellular membrane, clearly showing that CD147 is a chaperone for its interacting partners including: MCT1, MCT4 PMCA1 and
ICAM1.
Furthermore, we tested the consequences of CD147 re-introduction into the downregulated cell lines to establish if the functional consequences of CD147 depletion can be restored with CD147 gene knock in (rescue). We generated stable CD147 knock-in cell lines by lentiviral transduction of shCTRL or shCD147 cells with empty vector (CTRL), wild type CD147 or a glycosylation incapable mutant (CD147NG), followed by selection with neomyocin. The CD147NG vector contained full-length CD147 with all three sites of glycosylation (N44, N152 and N186) mutated to glutamine to prevent glycosylation. This glycosylation incapable mutant allowed us to test the role of glycans in CD147 chaperone
62
activity. This is especially significant, since the importance of CD147 glycosylation on its activity remains controversial.
Figure 35A shows that stable introduction of two different CD147 constructs led to re- expression of CD147 in shCD147 cells, although not to a full extent as compared to the shCTRL cells transduced with empty lentivector. In addition, there was a significant reduction in CD147NG molecular weight consistent with the loss of glycosylation in those cells. This restoration of CD147 expression was partially accompanied by an increase in cell growth (Figure 35B), with statistical significance determined only for CD147NG cells. This is most likely due to only partial re-expression of CD147 in the knockdown cells. We next evaluated the cellular localization of CD147NG to determine if this non-glycosylated form of
CD147 translocates to the membrane, as we would expect for the wild type CD147. As it can be seen in Figure 35B, CD147NG exhibited the same localization pattern as the wild type protein, also confirmed by transfections with C-terminal GFP fusion constructs (Figure 35C bottom panel). This observation is consistent with the above results showing that the glycosylated CD147-ECD is only weakly active without a significant contribution from the glycans (Figure 19). Based on our results depicting that both wild type and unglycosylated
CD147 are properly localized in PDAC cells, we proceeded to test the effect of their re- expression in regard to CD147 interactions and activity. Western blot analysis revealed that restitution of CD147 expression is consistent with rescue of its interacting protein expression
(Figure 35D). More importantly, our CD147 rescue cell lines restored the intracellular lactate levels (Figure 35E), which is an indication of the return of MCTs activity. Furthermore, the level of CD147-ECD glycosylation did not lead to any significant differences between the glycosylated versus unglycosylated CD147 activity, as indicated by intracellular lactate levels and interacting proteins expression. These observations further support our results suggesting that CD147 regulates PDAC progression via an interaction with small molecule transporters on the cell membrane.
63
CD147 cellular interactions lead to the reprogramming of glucose and glutamine metabolism
Our SILAC and cross-linking/MS analysis suggest that CD147 regulates metabolic processes through its chaperone function. Furthermore, therapeutically targeting metabolic pathways has become a very attractive strategy directed at halting cancer progression.
Thus, we evaluated the exact metabolic changes taking place in shCD147 cells. We have already shown that CD147 depletion in PDAC cell lines led to an accumulation of intracellular lactate (Figure 27), however, as illustrated by global metabolomic analysis, the inhibition of MCTs expression (MCT4) and activity (MCT4 and MCT1) did not recapitulate the same overall metabolome profile as in shCD147 cells (Figure 28). To directly characterize the extent of CD147 signaling in regard to metabolic changes in PDAC cells,
13 13 15 we used isotope tracers, C6-glucose and C6, N2-glutamine, to assess the metabolic flux by monitoring the incorporation of the heavy carbon and nitrogen isotopes into metabolites.
We performed labeling experiments in PANC1 cells, since these cells exhibited the highest level of phenotypic changes upon CD147 depletion. To account for MCT-driven processes, we also utilized MCT4 depleted cells. We did not test MCT4 cells treated with MCT1inh in here since the global metabolomic changes of those cells were similar to those of shMCT4 cells (Figure 28). We grew control (shCTRL) or knockdown (shCD147) PANC1 cells in the
13 13 15 presence of uniformly labeled C6-glucose or C6, N2-glutamine and then measured the incorporation of labeled molecules into specific metabolites and pathways. We performed those experiments in a time dependent manner and obtained time dependent responses but for simplicity only 1-hour time point and 24-hours time point for glucose and glutamine, respectively, is shown in the following figures (Figure 36 and Figure 38).
Cancer cells rely heavily on glucose uptake to supply cells with energy for growth and invasion. Cancerous cells take up large amounts of glucose by the action of membrane
64
bound glucose transporters (GLUTs). GLUT1 (gene name: SLC2A1) in particular, is highly expressed in glycolytic tumors and was one of the proteins identified in SILAC experiments to be highly upregulated in shCD147 cells (Figure 24). To determine if GLUT1 overexpression is also associated with increased glucose uptake, we measured the consumption of glucose upon CD147 or MCT4 depletion. Consistent with the increased expression of glucose transporter, we detected elevated levels of labeled glucose (M+6) in
CD147 depleted cells, with no difference in unlabeled intracellular glucose (Figure 36A).
Interestingly, shMCT4 cells imported less labeled glucose and had lower total levels of glucose than the control line. In each case we also observed increased labeling of other glycolytic intermediates (Figure 36A and B, glucose-6-phosphate, G6P shown), suggesting that both knockdown cell lines upregulate glycolysis, though they might rely on different carbon sources for this process. We also observed significant decrease in lactate export in each cell line relative to their control, as indicated by a ratio of labeled (M+3) extracellular lactate to labeled intracellular specie, and this was consistent with increased levels of intracellular lactate isotopologue (M+3). This data implies that although both CD147 and
MCT4 knockdown cells display an increase in glycolysis, CD147 depleted cells are less capable of removing lactate. Our data explains this, as CD147 knockdown results in the reduction of both MCT4 and MCT1, while that of MCT4 knockdown only effects MCT4 cellular expression.
Since accumulation of cellular lactate can lead to a buildup of glycolytic intermediates, we also tested the possibility of increased activity of other metabolic pathways. For example, G6P (M+6) and glucose-3-phospate (G3P, M+3) both increased in shCD147 cells (Figure 36A and B), directly linking glycolysis with the pentose phosphate pathway (PPP). Therefore, we next tested the activity of PPP by assessing the labeling pattern of 6-phosphogluconate (6PG, M+6) and pentose phosphates (PP, M+5), which are produced in this pathway. We observed large accumulation of those metabolites, suggesting
65
upregulation of the PPP pathway (Figure 36B). The increase in PPP was significantly more robust in CD147 depleted cells with only modest rise in MCT4 depleted cells.
This substantial upregulation of glycolysis and the PPP, along with a build up of lactate, would suggest that the produced metabolites are feeding into other pathways. Thus, we next analyzed TCA cycle intermediates labeling to account for any compensatory activity of this pathway. We evaluated the levels and labeling pattern of citrate (M+2) and αKG
(M+2), which are both indicative of the rate of citrate and αKG synthesis and their downstream utilization. Citrate can be generated from glutamine via reductive carboxylation of glutamine-derived-αKG, and so an increase in αKG/citrate ratio could be indicative of increased glutamine consumption to fuel TCA cycle. Another way of citrate generation is from forward reactions of the TCA cycle, where pyruvate-derived Acetyl-CoA, produced from de-carboxylation of M+3 pyruvate, provides building blocks for production of M+2 citrate isotopologue. This process leads to an increase in oxidative glutamine carboxylation, which in aerobic cells can correlate with increased fatty acid production80. Interestingly, we detected different levels of these metabolites in CD147, as opposed to MCT4 depleted cells
(Figure 36C), once again indicating that CD147 PDAC regulation employs a distinct mechanism that differs from MCT4 mediated signaling. shMCT4 cells exhibited a buildup of
αKG, while shCD147 cells contained much higher levels of citrate. Consistent with increased glucose consumption presented in Figure 36A, the αKG/citrate ratio along with an increased
M+2 isotopologue suggests that CD147 depleted cells rely heavily on glucose for citrate production and on glutamine for oxidative glutamine metabolism that fuels TCA cycle.
To further confirm the presence of oxidative glutamine metabolism in CD147 depleted cells, we analyzed cell extracts and media from cells cultured in the presence of
13 15 isotopically labeled glutamine ( C6, N2-glutamine). We uncovered an increased level of glutamine labeling in shCD147 cells, indicating an increased consumption of glutamine
(Figure 36D). This was in agreement with elevated expression of ASCT2, which is a
66
glutamine and neutral amino acid transporter (Figure 37D). We next evaluated the ratio of oxidative carboxylation derived citrate (M+4) to reductive carboxylation derived citrate (M+5) and, as our data indicates (Figure 36D), CD147 knockdown cells confide to oxidative carboxylation. Interestingly, we also detected large amounts of different glutamine isotopologues in shCD147 cells media (Figure 36D), which implies that while glucose is utilized for glycolysis and the PPP pathway, CD147 depleted cells rely more heavily on glutamine for fueling anaplerotic pathways.
CD147 mediated growth inhibition is linked to metabolic reprogramming
Our metabolomics analysis revealed that depletion of CD147 in PDAC leads to a striking metabolic reprogramming with a subsequent escalation of oxidative carboxylation.
An increase in oxidative carboxylation typically leads to an increase in fatty acid synthesis, which can provide building blocks for cell growth. Although we were unable to identify heavy carbon incorporation into lipids, our functional data suggests that impediment in cell proliferation of shCD147 cells (Figure 21) is consistent with an increase in their size (Figure
37A), pointing to the possible enhanced production of fatty acids for cell membrane expansion. Additionally, we detected cell cycle arrest in S/G2 phase (Figure 37B) concomitant with an augmented cellular DNA content, which could also aid to an increase in cell size. This increase in size along with cell cycle arrest and decrease in proliferation, could possibly link CD147 metabolic reprograming to the observed phenotypes. Additionally, cell cycle arrest is consistent with deregulated integrin signaling observed in here (Figure
24B and Figure 37D). Integrin signaling is one of the regulatory pathways important for maintenance of cell cycle checkpoints156.
The accumulation of metabolic intermediates can lead to an increase in cellular toxicity and cell death. Therefore we measured cellular apoptosis via Annexin V staining and, surprisingly, we did not observe a significant increase in cell death due to the loss of
67
CD147 (Figure 37C). The lack of cell apoptosis in CD147 depleted cells suggests compensatory mechanism for dealing with buildup of lactate and glycolytic intermediates, i.e., increase in size. Interestingly, one of the common trends that we detected in our SILAC and cross-linking/MS analysis was deregulation of cellular transporters and enzymes involved in amino acid synthesis (Figure 24). Global metabolomics analysis revealed upregulation of several neutral amino acids (Figure 28), which we further confirmed with metabolic tracing analysis. The use of glucose tracer revealed a significant rise in intracellular labeled alanine (M+3), proline (M+2) and glycine (M+2) levels in shCD147 cells
(Figure 38A). Although we were unable to detect labeled proline in the media, we measured high levels of alanine isotopologue (M+3) and glycine (M+0+1) suggesting not only an increased production of this amino acid but also its increased secretion. Importantly, alanine is produced from the transamination of pyruvate. Therefore, excess lactate could allosterically inhibit the forward reaction of lactate dehydrogenase, thereby resulting in the accumulation of pyruvate, which would then act as a substrate for alanine aminotransferase
(ALT). Simply put, elevated alanine production/export in shCD147 cells as demonstrated by our steady state and flux-balance metabolomics analyses represents a mechanism for the cells to adapt to increased intracellular lactate and indeed, expression levels for ASCT2 are increased in these cells.
We also evaluated the contribution from glutamine to the production of the amino acids described above. This is especially important since glutamine, via its involvement in transamination reactions is a nitrogen donor for the formation of amino acids. Furthermore, proline is almost exclusively made from glutamine and its intermediates such as glutamate, and monitoring proline labeling is an accurate readout of glutamine utilization. Indeed, we detected full labeling for both extracellular and intracellular proline (M+5+1) as well as significant increase in total intracellular proline (Figure 38B). Likewise, glycine and alanine each contained glutamine derived-nitrogen (M+0+1) and their levels were increased in
68
shCD147 cells (Figure 38B). This data is consistent with an increase in amino acid transporters expression in CD147 depleted cells (Figure 37D). Remarkably, glycine levels were only slightly increased in the media suggesting that glycine might be involved in other intracellular process. This is particularly important since we also detected increased levels of phosphoglycerate dehydrogenase (PHGDH) an enzyme, which catalyzes a rate-limiting step in serine biosynthesis. Serine biosynthesis is a prerequisite for glycine generation and subsequent nucleotide synthesis. Though we were not able to detect serine in our experiments we observed elevated expression of several nucleotides. As shown in Figure
38B, shCD147 cells produce large amounts of nucleotides (representative thymine is shown) supporting the supposition that shCD147 cells remove glycolytic intermediates by producing building blocks for cell growth. However, due to cell cycle arrest, CD147 knockdown cells are unable to divide and instead they utilize the produced building blocks for cell enlargement. Taken together our metabolic analysis reveals that CD147 loss guides cells to metabolic reprograming and activation of compensatory pathways to counteract the detrimental effects of CD147 depletion.
Discussion
The role of CD147 in regulating cancer progression has been broadly studied thus far, but the molecular details remain elusive. The work described in this study is the first comprehensive analysis that has identified the specific CD147 interacting partners and molecular details of its function, which has revealed a chaperone role in regard to regulating
PDAC progression. Our analysis discovered that CD147 plays a chaperone function for multiple transmembrane proteins, whereby CD147 both protects its target proteins from degradation and trans-locates these proteins to the cellular membrane. In addition, we characterized the role of CD147 glycans in regard to the activity and interactions of the extracellular region and the full length protein.
69
We identified that CD147 glycosylation does not play an important role in CD147 activity, an observation previously discarded. This is an important discovery, since the specific contribution of CD147 glycans to its activity have been inconclusive. In particular, previous studies show that glycans are required for CD147 membrane localization, however our studies did not recapitulate these results94. It is possible that depending on the disease state and cell type, CD147 exhibits varying glycosylation needs and in PDAC cells glycans do not play an important role in regard to CD147 activity. Consistent with the impact of glycans on full length CD147 expression and function, we also did not observe any specific activity for the CD147 extracellular region in regard to its glycosylation and interactions. It is conceivable that the purified CD147-ECDgly did not contain full extent of glycosylation, or glycosylation pattern differs between different cell lines. Such lack of correct glycosylation might lead to inadequate conclusions in regard to CD147-ECDgly activity, and might explain the conflicting results reported in literature. It is also feasible that glycans are important for
CD147 mediated cell-cell communication, but are not dispensable for its other functions, not studied here. Based on our analysis, the role of CD147 in regulating PDAC progression is likely independent of its glycosylation status.
Our studies further exemplified the importance of CD147 in regulating cancer progression. We uncovered that independently of the initial cancer phenotype (i.e. mesenchymal versus epithelial cells), the loss of CD147 led to an attenuation of cell growth and proliferation. Although we observed the highest level of growth suppression in most mesenchymal cells, the loss of CD147 affected cell proliferation across all cell lines tested.
The appearance of the most significant growth attenuation in mesenchymal cells suggests
CD147 mediated regulation of the EMT, a process implicated in the progression of cancer growth, metastasis and drug resistance73, 140. Recently Sinha et al., performed a comprehensive genomic study in 26 PDAC cell lines, including PANC1 and MiaPaCa2, and identified a diverse set of genomic alterations (deletions and mutations) in the mesenchymal
70
cells as opposed to the epithelial cells157. The appearance of those genetic alterations could explain the striking consequences of CD147 depletion on mesenchymal-like cells and the subsequent EMT reversal. It is possible that CD147 functions in a manner dependent on specific gene composition, that alters between those two cellular phenotypes. Furthermore, previous studies proposed a mechanism in which CD147 interaction with CD44 regulates the EMT via activating pSTAT3 signaling158. Consistent with this known pathway our SILAC analysis revealed significant downregulation of CD44 in shCD147 cells however, we did not detect any changes in STAT3 phosphorylation status (data not shown) aiming to the supposition that CD147 regulation of the EMT in PDAC cells is not via the pSTAT3 signaling, but could still be CD44 dependent considering the decreased expression of this protein. Previous studies have also shown a possible interaction between CD44 and CD147 however, we only observed a very modest enrichment of CD44 in cross-linking/MS data, which we were not able to further confirm by other methods112, 113. Another possibility of the
EMT regulation is EGFR mediated signaling. Indeed, EGFR has been extensively characterized in regard to EMT and EGFR/CD44 mediated signaling in the EMT progression has also been reported112. Although EGFR was one of the proteins identified in our cross- linking/MS assay, we were unable to further confirm an interaction between this protein and
CD147. However, the observed reversal of the EMT could be explained by the altered signaling of EGFR and CD44 detected in CD147 depleted cells, as opposed to the direct contribution of specific interactions.
We also established that although CD147 and MCTs are tightly linked together, their signaling requires different cofactors and leads to different phenotypic changes. This is an important point for future development of MCTs or CD147 targeting strategies in regard to
PDAC progression. Specifically, CD147 downregulation led to a significant attenuation of
MCT4 expression, but an independent loss of MCT4 expression had different functional consequences on the cells. Our analysis suggests that there is a cross-talk between MCTs
71
and CD147 effects on cellular metabolism, however, CD147 requires additional factors not influenced by MCT4 depletion. For example, our studies along with previous analysis indicate that MCT4 downregulation does not affect the EMT in PDAC cells, although cell growth is slowed in each case129. We have further showed that the regulation of cell progression in PDAC cells, upon CD147 knockdown, is due to the reversal of EMT and sequestration of β-catenin to the membrane. This mechanism of cancer inhibition appears to be independent of MCT4 expression and points to the contribution of other proteins under
CD147 regulation, which may also be chaperone targets of CD147 like the proteins we have identified here.
We performed a comprehensive SILAC-based proteomics analysis, in order to first identify the specific proteins deregulated due to the loss of CD147 expression. We then followed this study with a cross-linking/MS analysis, in which we detected several possible
CD147 interacting partners. Interestingly, both experiments displayed significant overlap in terms of the functional pathways classified. In each case numerous proteins involved in the regulation of cellular metabolism and adhesion were detected. This cross-talk between the two experiments suggests that CD147 interacts with some of the identified proteins, which we have confirmed through multiple experiments, leading to direct influence on their activity.
One such case would be an interaction with MCT4, where the downregulation of CD147 was followed by accumulation of lactate. Another example is interaction with ICAM1, which plays an important role in cell migration and metastasis. ICAM1 is typically upregulated in PDAC, where it plays a critical role in mediating inter-cellular communication vital for cell migration159. CD147 mediated downregulation of ICAM1 advocates a direct role for CD147 in controlling cell migration, via targeting ICAM1 to the membrane. It is also feasible that our identified PMCA1 interaction with CD147 is a direct link to the regulation of the EMT and cell proliferation in PDAC cells, since tainted Ca signaling has been implicated in regulation of those processes160. PMCA1 is important for maintaining both intracellular and extracellular
72
levels of Ca ions. Furthermore, increased extracellular levels of Ca are associated with augmented E-cadherin expression and increased cell-cell adhesion. Although we did not directly test calcium levels in our CD147 depleted cells, Supper et al., recently showed that
CD147 can interact with PMCA4 in Jurkat T cells, and CD147 downregulation alters calcium mobilization dynamics in those cells152. Thus, it is reasonable to propose that altered
PMCA1 expression in PDAC cells could lead to altered Ca levels and thus influence E- cadherin expression and signaling.
Our data strongly supports a protective role for CD147 in terms of its interacting partners expression, since in the absence of CD147 these proteins are targeted for degradation. Our analysis revealed that CD147 interacts with multiple proteins and acts as a chaperone for those proteins. Chaperones play a critical role in de novo protein folding of nascent polypeptides, protection from degradation, protein quality control regulation and translocation across membranes161, 162. Although we did not test the contribution of CD147 expression to nascent peptide folding, the lack of regulation of this process by other chaperones has also been described. Specifically, Hsp90 a member of the heat shock family (HSPs) is a ubiquitously expressed chaperone protein not involved in de novo protein folding, but instead implicated in the interactions with signal transduction factors163. Similarly to CD147, Hsp90 is overexpressed in multitude of cancers, where it interacts with numerous
GPCRs and this interaction is critical for proper expression and localization of those receptors163. Interestingly, a novel membrane bound form of Hsp90 (Hsp90N) has been recently identified, important for targeting Raf to the plasma membrane to trigger its functional activity164. This Hsp90N activity is in agreement with our observed results, in which CD147 targets transmembrane proteins to the cellular membrane allowing them to function properly, as indicated by MCT4 activity (Figure 35E).
Molecular chaperones typically function in a large complex with other co-chaperone proteins. For example, HSPs form multi-subunit structures, where different accessory
73
proteins play diverse functions and, thus, chaperone interaction networks are typically composed of high number of proteins163. Calnexin, an ER membrane bound chaperone important for glycoprotein biogenesis, forms large complexes with its interacting partners and other co-chaperones (celreticulin)165. Consistent with these phenomena, our cross- linking/MS analysis identified a multitude of CD147 interactions, including known chaperones and secretory proteins (i.e., calnexin, calreticulin, Syntaxin 4; see Table 4 for more examples). This data indicates that CD147 forms a large complex with multiple proteins and requires other co-chaperones for proper membrane targeting. In this regard, targeting CD147 might have a more drastic effect in blocking cancer progression than specifically targeting its interacting partners (e.g. MCT4, MCT1, ICAM1, PMCA1).
The development of therapeutic strategies entails a comprehensive understanding of molecular mechanisms therefore, we aimed to acquire a detailed analysis of CD147 signaling cascades. To this end, steady state and metabolic flux experiments allowed for deeper examination of specific processes under CD147 regulation. Downregulation of
CD147 expression results in the downregulation of the EMT and β-catenin sequestration to the membrane, which suppresses β-catenin transcriptional activity. Further, this reprogrammed signaling suppressed cell growth, which was concomitant with a modest increase in cell size and cell cycle arrest. In addition, CD147 depletion led to possible escalation of glucose and glutamine uptake. This finding was unexpected, given that the consumption of both of those carbon sources is already enhanced in PDAC cells. Glucose and glutamine labeling experiments revealed that CD147 downregulation increased the generation of cell building blocks (amino acids and nucleotides). This data is consistent with an increase in cell size and implies a compensatory mechanism for the removal of metabolic intermediates. The arrest in S/G2 prevents cells form dividing, forcing them to increase their cell mass by accommodating the enhancement in glycolytic intermediates generation. It is likely that due to the deregulation of several proteins, upon CD147 depletion, cells are
74
compensating for a feedback loop, in which pathways feed into each other to maintain cellular homeostasis. Our metabolomic analysis uncovered that CD147 depletion leads to reprogramming of PDAC cell metabolism, in which cells continue to rely heavily on glucose and glutamine, but alter their glutamine metabolism by activating oxidative glutamine carboxylation. This unique adaptation to intracellular lactate accumulation appears to be an efficient mechanism to protect cells from apoptotic cell death. Indeed, loss of CD147 expression did not affect cell viability while the depletion of MCT4 parallels significant cell death in PDAC129.
CD147 halts PDAC cell growth by arresting cells in S/G2 phase. Our SILAC-based
MS analysis reveled altered integrin signaling in PANC1 cells, which could explain the distortion in cell cycle and proliferation in those cells. Cell cycle progression can be regulated through integrin signaling, where integrin mediated interactions with the ECM and receptor binding of growth factors alters cells entry into different stages of the cell cycle156.
Interestingly, Keremer et al. discovered that integrin β4, via its interaction with integrin α3 and laminin 5 can attenuate cell cycle arrest in G2 phase by increasing cellular adhesion in prostate cancer166. Given that we observed increased adhesion and integrin β4 expression, this integrin mediated cell cycle regulation could be one explanation for the observed S/G2 arrest in CD147 deleted cells. Furthermore, our data strongly suggests that CD147, via regulating the expression and localization of membrane proteins indirectly regulates several signaling events, including cell cycle arrest, β-catenin translocation, EMT and metabolism.
Although, we have shown an interaction between CD147 and a subset of membrane proteins (MCT4, MCT1, ICAM1, ASCT2 and PMCA1) our cross-linking/MS analysis identified several other proteins that could interact with CD147, but were not further characterized in here. Our observation that some of the proteins, identified to interact with
CD147 were upregulated due to the loss of CD147 expression, also insinuates that they might form a large complex with CD147, rather then directly interact. Based on an
75
increased production of their ligands (i.e. amino acids for ASCT2, LAT1 and CD98hc), we reasoned that CD147 depletion activates a compensatory loop where downregulation of its direct interacting partners activates synchronous upregulation of some of its indirect partners, to counteract the consequences of CD147 depletion. Taken together our data provides rationale for developing selective impasse strategies to inhibit CD147 activity in
PDAC progression.
Future directions
The studies described here provide a comprehensive analysis of CD147 function in regard to PDAC reprogramming due to CD147 deregulation. Although, our analysis illustrates that CD147 interacts with several transmembrane proteins and the loss of these interactions leads to striking cellular reprogramming, several unanswered questions remain.
For example, what are the specific regions of CD147 responsible for the interactions? We have shown that the glycosylation status does not play an important role in mediating
CD147 engagement with other proteins, thereby suggesting that the Ig-like domains of
CD147 do not play a role in such interactions as previously shown95. However, we did not test the contribution of the cytoplasmic tail or transmembrane region. The latter might be especially important, since the TM region contains a conserved glutamic acid in the middle of its membrane-spanning domain. Such localization of a charged residue is quite unusual for membrane proteins and would suggest a specific functional consequence. Thus, future studies will be critical in testing the role(s) of both the CD147 TM and C-terminal domains to determine whether there are specific interactions for different CD147 chaperone targets or whether these interactions utilize the same CD147 regions.
76
Table 4 Proteins identified in cross-linking experiments in PANC1 cells
Identified proteins* Exp#1 unique peptides Exp#2 unique peptides CD147- CD147- Gene Accession CD147- GFP + CD147- GFP + Symbol Number GFP GFP GFP+DSS DSS GFP GFP GFP+DSS DSS ATP1A1# P05023 0 0 0 28 0 0 0 43 ABCC1 P33527 0 0 1 45 0 1 1 51 AT2B1 P20020 0 0 0 28 0 1 0 34 SLC16A3 O15427 0 6 0 7 2 7 0 8 PTRF Q6NZI2 0 0 1 15 0 0 2 16 SLC7A5# Q01650 0 1 1 6 0 1 2 7 TFRC P02786 0 2 1 25 0 2 2 24 ITGB1 P05556 0 0 2 19 0 2 2 24 SLC1A5# Q15758 1 2 2 11 2 2 2 12 HM13 Q8TCT9 0 1 0 8 0 0 0 9 MYADM Q96S97 0 1 1 5 0 2 2 6 PLXNB2 O15031 0 0 0 22 0 0 0 26 EGFR# P00533 0 1 2 26 1 3 2 28 ITGA3 P26006 0 0 0 14 0 0 0 19 ITGA2 P17301 0 0 0 21 0 0 1 21 CDCP1 Q9H5V8 0 6 0 9 0 13 0 12 TECR Q9NZ01 1 2 2 5 2 2 2 5 CALR P27797 0 2 0 9 0 0 1 10 CAV1# Q03135 1 3 2 5 1 5 1 6 HLA-A# P13746 1 2 0 7 1 4 1 9 RAP1B P61224 0 0 2 8 0 0 2 8 AHCYL1 O43865 0 0 1 12 0 0 1 12 DDOST P39656 0 3 2 8 1 1 1 7 PLXNA1 Q9UIW2 0 0 0 14 0 0 0 23 SEL1L# Q9UBV2 0 0 0 14 0 0 0 12 SLC12A7 Q9Y666 0 0 0 12 0 0 0 17 SLC29A1 Q99808 0 0 0 8 0 0 0 8 SLC26A2 P50443 0 0 0 7 0 0 0 10 SLC4A7 Q9Y6M7 0 0 0 13 0 0 1 18 L1CAM P32004 0 0 0 17 0 0 0 15 LGALS8 O00214 0 6 0 8 0 7 0 10 MXRA8 Q9BRK3 0 0 0 11 0 1 0 13 MET P08581 0 1 0 8 0 7 0 17 GNAI2 P04899 0 1 1 9 1 1 1 9 STT3A P46977 0 2 2 9 1 0 0 4 CD97 P48960 0 0 0 10 0 1 0 13 SUSD2 Q9UGT4 0 0 0 9 0 0 0 11 PTK7 Q13308 0 0 0 15 0 0 0 21 SLC16A2# P36021 0 0 0 4 0 1 0 3 SLC4A2 P04920 0 0 0 15 0 0 0 25 MPP6 Q9NZW5 0 0 0 9 0 0 0 9 RPS27 P42677 2 2 2 3 1 3 1 3 PRKCSH P14314 0 0 0 11 0 0 1 12 ACSL4 O60488 0 2 1 7 0 2 1 10 NT5E P21589 0 0 0 8 0 5 0 13 EPHB4 P54760 0 0 0 13 0 0 0 24 CD151 P48509 0 0 0 6 0 0 0 6 ADAM10 O14672 0 0 0 12 0 0 0 12 ASPH Q12797 0 0 0 12 0 1 0 3 SLC16A1# P53985 0 0 0 5 0 0 0 5 SLC30A1 Q9Y6M5 0 0 0 10 0 0 0 10 TMED9 Q9BVK6 0 1 0 5 0 1 0 4 ACSL3 O95573 0 1 2 8 1 4 1 5 LMAN1 P49257 0 1 0 7 0 0 0 3 CDC42 P60953 0 0 1 4 1 0 1 6 SPCS2 Q15005 0 1 1 4 0 1 0 3
77
Table 4 Continued Identified proteins* Exp#1 unique peptides Exp#2 unique peptides CD147- CD147- Gene Accession CD147- GFP + CD147- GFP + Symbol Number GFP GFP GFP+DSS DSS GFP GFP GFP+DSS DSS STT3B Q8TCJ2 0 2 0 10 0 0 0 4 TM9SF2 Q99805 0 0 0 6 0 0 0 7 SCAMP1 O15126 0 0 0 5 0 0 0 9 ITGAV P06756 0 0 0 15 0 0 0 9 SCAMP3 O14828 0 0 0 4 0 2 1 4 TMX1 Q9H3N1 0 0 0 5 0 1 0 7 IGF2R P11717 0 0 1 13 0 0 0 7 RHOA P61586 1 0 0 5 2 2 2 5 CD59 P13987 0 0 0 4 0 2 1 4 ERLEC1 Q96DZ1 0 0 0 8 0 0 0 6 PIGS Q96S52 0 0 0 9 0 0 0 6 NOMO2 Q5JPE7 0 0 0 15 0 0 0 3 CD109 Q6YHK3 0 0 0 9 0 0 0 13 NF2 P35240 0 0 1 6 0 0 2 4 ATP2B4 P23634 0 0 0 9 0 0 0 10 ITGA5 P08648 0 0 0 10 0 0 0 13 RAB14 P61106 1 0 2 4 1 0 0 5 POR P16435 0 0 2 9 0 0 1 8 STXBP3 O00186 0 0 0 9 2 3 0 5 EHD4 Q9H223 0 0 1 4 0 1 1 8 EMC1 Q8N766 0 0 0 9 0 0 0 3 RAB2A P61019 2 1 2 3 1 4 1 6 ATP1B1 P05026 0 0 0 4 0 0 0 3 SACM1L Q9NTJ5 0 0 0 9 0 0 0 8 TAP2 Q03519 0 2 0 5 1 1 1 3 ATP1B3# P54709 0 0 0 4 0 0 0 8 NRP1 O14786 0 0 0 10 0 0 0 12 CNNM3 Q8NE01 0 0 0 4 0 0 0 8 SLC20A1 Q8WUM9 0 0 0 10 0 0 0 6 ANO6 Q4KMQ2 0 0 0 5 0 0 0 11 LMAN2 Q12907 0 0 0 5 0 0 0 7 PVRL2 Q92692 0 0 0 7 0 0 0 9 SCRIB Q14160 0 0 0 8 0 0 0 14 LPCAT1 Q8NF37 0 1 2 4 1 3 0 4 EEPD1 Q7L9B9 0 0 0 4 0 0 0 9 UGGT1 Q9NYU2 0 0 0 8 1 0 0 5 ITGB4 P16144 0 0 0 8 0 0 0 6 TBL2 Q9Y4P3 0 2 0 4 1 0 1 4 CD70 P32970 0 0 0 6 0 0 0 6 F11R Q9Y624 0 0 0 7 0 0 0 8 APOL2 Q9BQE5 0 0 0 5 0 0 1 3 SLC9A3R2 Q15599 0 1 0 5 0 0 0 7 RAC1 P63000 0 0 0 5 0 0 1 5 TAP1 Q03518 0 2 0 4 1 0 0 4 ITFG3 Q9H0X4 0 0 0 4 0 0 0 8 CD99L2 Q8TCZ2 0 0 0 3 0 0 0 4 TMEM43 Q9BTV4 0 0 0 6 0 0 0 7 SLCO4A1 Q96BD0 0 0 0 5 0 0 0 6 SLC1A4 P43007 0 0 0 5 0 0 0 8 SLC5A3 P53794 0 0 0 5 0 0 0 6 DLG1 Q12959 0 0 0 6 0 0 0 5 F2RL1 P55085 0 0 0 5 0 0 0 5 SLC19A1 P41440 0 0 0 5 0 0 0 6 GNAI3 P08754 0 0 1 5 0 0 0 5 PIGK Q92643 0 1 0 5 0 0 0 3 APMAP Q9HDC9 0 0 0 4 0 0 0 4
78
Table 4 Continued Identified proteins* Exp#1 unique peptides Exp#2 unique peptides CD147- CD147- Gene Accession CD147- GFP + CD147- GFP + Symbol Number GFP GFP GFP+DSS DSS GFP GFP GFP+DSS DSS HMOX2 P30519 0 0 0 4 0 0 0 4 SURF4 O15260 0 0 1 5 0 0 1 3 TPBG Q13641 0 0 0 3 0 0 0 6 STEAP3 Q658P3 0 0 0 4 0 0 0 5 ITGA6 P23229 0 0 0 7 0 0 0 7 ADAM9 Q13443 0 0 0 4 0 0 0 8 PGRMC1 O00264 0 0 0 4 0 0 0 4 ACVR1 Q04771 0 0 0 5 0 0 0 4 LDLR P01130 0 0 0 7 0 0 0 7 ABCC3 O15438 0 0 0 6 0 0 0 7 ALCAM Q13740 0 0 0 5 0 0 0 8 STX4 Q12846 0 0 0 4 0 0 0 5 EHD2 Q9NZN4 0 0 2 4 0 0 0 7 EPHX1 P07099 0 0 0 5 0 0 0 3 SLC12A2 P55011 0 0 0 7 0 0 0 7 PCDH7 O60245 0 0 0 8 0 0 0 6 MPZL1 O95297 0 0 0 3 0 0 0 4 ZDHHC5 Q9C0B5 0 0 0 4 0 0 0 4 CD46 P15529 0 0 0 3 0 0 0 3 JAK1 P23458 0 0 0 7 0 0 0 4 CLDN4 O14493 0 0 0 3 0 0 0 4 EPHB2 P29323 1 0 0 4 0 0 0 5 ERGIC1 Q969X5 0 0 1 4 0 1 1 3 COLGALT1 Q8NBJ5 0 0 1 4 1 0 0 3 EMC2 Q15006 0 0 1 4 0 0 1 4 TMX2 Q9Y320 0 0 0 4 0 0 0 3 SLC35F2 Q8IXU6 0 0 0 3 0 0 0 4 GLIPR1 P48060 0 0 0 3 0 0 0 6 PODXL O00592 0 0 0 3 0 0 0 5 LRRC8C Q8TDW0 0 0 0 4 0 0 0 3 SCAMP2 O15127 0 0 0 3 0 0 0 3 SLC7A1 P30825 0 0 0 4 0 0 0 5 DISP2 A7MBM2 0 0 0 4 0 0 0 3 NPTN Q9Y639 0 0 0 3 0 0 0 3 SLC5A6 Q9Y289 0 0 0 3 0 0 0 4 TMEM30A Q9NV96 0 0 0 5 0 0 0 3 TSPAN15 O95858 0 0 0 3 0 0 0 3 LSR Q86X29 0 0 0 4 0 0 0 4 DCBLD1 Q8N8Z6 0 0 0 4 0 0 0 3 SLC3A2 #,** P08195 1 13 2 25 9 20 4 27 ICAM1** P05362 0 1 3 18 2 5 5 21 *Only proteins that met the selection criteria are listed. Out of 1469 proteins 153 were identified with at least 3 unique peptides and were present in CD147-GFP ± DSS samples in both biological replicate (Exp#1 and Exp#2). **Proteins selected for further analysis that were enriched in CD147-GFP samples but did not fulfill the selection criteria. See text for details. #Proteins previously reported in literature to interact with CD147. Discussed in CD147 interacting partners section.
79
Table 5 Inter-protein cross-linked peptides identification using Protein Prospector from cross-linked samples
CD147 Identified cross-linked protein Gene Peptide Residue Peptide Residue symbol SSEHINEGETAMLVCK*SESVPPVTDWAWYK 127 NOMO2 AK*PGTYK 458 VK*AVK 128 CD59 AGLQVYNK*CWK 63 GGVVLK*EDALPGQK 63 JAK1 NK*NK 972 SSEHINEGETAMLVCK*SESVPPVTDWAWYK 127 JAK1 FK*VAK 693 K*PEDVLDDDDAGSAPLK 234 AT2B1 CFK*ILSANGEAK 613 K*PEDVLDDDDAGSAPLK 234 EMC2 YFAQALK*LNNR 220 RK*PEDVLDDDDAGSAPLK 234 STXB2 TEQDLALGTDAEGQK*VK 382 RK*PEDVLDDDDAGSAPLK 234 SCRIB K*LGLSDNEIQR 63 GGVVLK*EDALPGQK 63 SCRIB K*LGLSDNEIQR 63
80
Figure 12 Graphical depiction of epithelial to mesenchymal transition (EMT) Epithelial to mesenchymal transition is a process by which epithelial cells loose their polarity and cell- cell adhesion and become migratory and highly invasive. E-cadherin and membrane bound β-catenin are markers of epithelial cells. Nuclear β-catenin (transcriptionally active) and Vimentin are markers of mesenchymal cells. EMT can be reversed to mesenchymal to epithelial transition (MET).
81
Figure 13 Metabolic reprograming in cancer cells Normal cells utilize glucose for anabolic glycolysis, which produces pyruvate. Pyruvate is fed into TCA cycle and oxidative phosphorylation. In normal cells this process produces 38 molecules of ATP for every molecule of glucose used, 2 ATPs from glycolysis and 36 ATPs from oxidative phosphorylation (OXPHOS). In addition, glutamine also supports the mitochondrial processes by fueling TCA cycle. In cancer cells glucose uptake is upregulated and subsequently glycolysis is increased. Metabolites produced in glycolysis are used to fuel other upregulated pathways including pentose phosphate pathway (PPP). Mitochondrial metabolic processes are decreased and glycolysis-derived pyruvate is converted to lactate and secreted outside of the cell. Glutamine uptake is also increased but an alternative pathway of glutaminolysis is activated. This metabolic reprograming leads to the production of 2 ATP molecules from each glucose molecule taken up by the cells. Pathways are bolded and underlined, and upregulated pathways are highlighted in red. Bolded arrows indicate increased flux.
82
Figure 14 Glutamine metabolism in cancer cells Cancer cells rewire their metabolic pathways by increasing their glycolytic flux to produce ATP. Glutamine metabolism is then redirected to produce citrate for fatty acid biosynthesis. Citrate can be produced via oxidative glutamine metabolism or reductive glutamine metabolism. Reductive glutamine metabolism supplies glutamine-derived carbons directly for citrate production.
83
Figure 15 Schematic representation of different CD147 isoforms and function (A) The two CD147 isoforms are depicted and only differ by the additional immunoglobulin (Ig) domain in isoform-1. The different topological regions are also indicated: ectodomian (ECD), transmembrane (TM) and cytoplasmic tail (CT). The conserved transmembrane glutamic acid (E218) is indicated with an arrow, and sites of glycosylation (N44, N152 and N186) are marked with a star. The glycosylation of Ig0 has not been fully characterized. Glycans are denoted by trapezoids. C and N-termini are designated. (B) The representative reported functions and interacting partners are depicted. The specific CD147 regions implicated in the functions or interactions are also marked.
84
Figure 16 The activity of CD147 ectodomain Extracellular region of CD147 (CD147-ECD) was purified from E. coli and used for stimulation of indicated pancreatic cancer cells. Cells were plated into 6-well plates and 24hrs later stimulated overnight with 50 µM CD147-ECD or buffer control, in serum free media. Cell supernatants were collected after the treatments and secretion of interleukin 6 (IL6) and matrix metalloproteinase 9 (MMP9) were measured using ELISA assay. Bars are ± SEM, n=4, ***p<0.001.
Figure 17 Transfection and purification scheme for obtaining pure glycosylated CD147 ectodomain Cells are transiently transfected on day 1 with construct encoding secretion peptide followed by purification tags, TEV cleavage site and CD147 residues 22 to 205. One day post transfection cells are treated with cell cycle inhibitor (2.2 mM valproic acid) to inhibit cell division and increase mRNA synthesis. Five days later conditioned media is collected and applied to a Ni-affinity column and CD147-ECDgly is eluted off the column, tags are cleaved with TEV protease and protein is further purified via size exclusion chromatography.
85
Figure 18 Purification of glycosylated CD147 Coomassie stained SDS-PAGE gel showing different stages of CD147-ECDgly expression (day 1-7) and elution from Ni-affinity and size exclusion (S75) columns. Last line shows PNGaseF treated purified CD147-ECDgly. PNGaseF removes N-linked glycans as indicated by lower molecular weight observed. MM - molecular marker.
Figure 19 The activity of glycosylated CD147 ectodomain measured by MMP9 ELISA Indicated cells were stimulated with increasing concentrations of glycosylated CD147 ectodomain (CD147-ECDgly) or buffer control for 24hrs in serum free media. Following treatment, conditioned media was collected and tested for MMP9 secretion via ELISA assay. Representative results are shown. Bars are ± SEM, n=2.
86
Figure 20 The Expression of CD147 and EMT markers in pancreatic cells The expression and glycosylation level of CD147 was evaluated by Western blotting with human CD147 specific antibody in indicated immortalized cell lines. The expression of E-cadherin and β-catenin (epithelial markers) and Vimentin (mesenchymal marker) was also analyzed. β-actin provided loading control.
Figure 21 Knockdown of CD147 in different PDAC cell lines Indicated cell lines were stably transduced with lentiviral particles encoding scramble control (shCTRL) or shRNA encoding different regions within CD147 gene. (A) CD147 downregulation was detected by Western blotting. β-actin provided loading control. (B) Cell growth was monitored by counting live cells on a daily basis for four days post seeding (day 1). All counts were normalized to day 1. Bars are ± SEM, n=3, ***p<0.001. Representative results are shown.
87
Figure 22 CD147 knockdown is associated with reversal of mesenchymal phenotype (A) The expression of EMT markers was tested in the indicated shCD147 and shCTRL cells via Western blotting. β-actin provided loading control (B) IF staining for indicated EMT markers in representative shCD147 cells for each indicated cell line. Scale bar 10 µm. Cells were counterstained with Hoechst (nuclei).
Figure 23 Migratory capabilities of CD147 depleted cells Cell migration was determined using Boyden chamber system. Equal amount of indicated cells was seeded in the upper chamber in serum free media and cells were allowed to migrate towards media containing serum, placed in the lower chamber. Measurements were normalized to the control cells (shCTRL) and represented as % migration. Bars are ± SEM, n=6, **p<0.01,***p<0.001.
88
Figure 24 SILAC MS analysis reveals several differentially regulated proteins in CD147 depleted cells
89
Figure 24 SILAC MS analysis reveals several differentially regulated proteins in CD147 depleted cells (A) Volcano plots representing ratio of CD147 depleted to control cells (log2 of shCD147/shCTRL) versus p values (-log10) of all proteins identified in 3 independent experiments from two different pancreatic cancer cell lines. Significantly upregulated proteins are marked in red and significantly downregulated proteins are marked in green. Significant deregulation was determined based on p<0.05 (ANOVA) and the appearance of the specific protein in at least two out of three independent biological replicates. (B) The significantly up or downregulated proteins are represented as a heat map, where the light colors correspond to the most significant changes. Gene names marked with stars were detected in both cell lines. (C) Gene ontology (GO) terms analysis for all statistically significant up or downregulated proteins from indicated cell lines. Calculation of over-represented GO terms was performed using the entire human genome as a background (threshold count = 2, EASE score = 0.1). Only terms with a p value < 0.05 are listed.
90
Figure 25 CD147 depletion leads to MCT1 and MCT4 downregulation The expression of MCT1 and MCT4 was measured in the indicated cell lines upon stable CD147 depletion via Western blotting. β-actin provided loading control.
91
Figure 26 MCT4 depletion in PDAC cell lines leads to CD147 downregulation but does not influence EMT Indicated cell lines were stably transduced with lentiviral particles encoding scramble control (shCTRL) or shRNA targeting MCT4 gene. (A) MCT4 and MCT1 downregulation was detected by Western blotting. β-actin provided loading control. (B) The expression of EMT markers (Vimentin, E-cadherin and β-Catenin) was measured in the indicated cell lines via Western blotting. β-actin provided loading control. (C) Cell growth was monitored by counting live cells on a daily basis for four days post seeding (day 1). All counts were normalized to day 1. Bars are ± SEM, n=3, ***p<0.001. Representative results are shown.
92
Figure 27 Intracellular lactate accumulation after CD147 and MCT4 depletion Metabolomics analysis for intracellular lactate levels. See Materials and Methods for experimental details. Bars are ± SEM, n=3, **p<0.01,***p<0.001.
93
relative
min max T4 T4 T4 hMC hMC hMC Pathway Metabolite shCD147 shCD147 shCD147 s s s shMCT4+MCT1inh shMCT4+MCT1inh shMCT4+MCT1inh Amino acids aspartate Amino acids arginine Amino acids asparagine Amino acids glycine Amino acids histidine Amino acids cysteine Amino acids proline Amino acids valine Amino acids glutamine Amino acids serine Amino acids lysine Amino acids alanine Amino acids threonine Amino acids leucine Amino acids phenylalanine Amino acids methionine Amino acids tyrosine Amino acids glutamate Amino acids tryptophan Amino acids cystine
Glycolysis D-Glucose Glycolysis D-Glucose 6-phosphate Glycolysis D-Glyceraldehyde 3-phosphate/Glycerone phosphate Glycolysis 2/3-Phospho-D-glycerate Glycolysis Lactate
TCA cycle Succinate TCA cycle Citrate TCA cycle Fumarate TCA cycle Malate TCA cycle Oxaloacetate
Pentose Phosphate Pathway 6-Phospho-D-gluconate Pentose Phosphate Pathway D-Erythrose 4-phosphate
GSH homeostasis Glutathione GSH homeostasis S-Glutathionyl-L-cysteine GSH homeostasis 5-Oxoproline GSH homeostasis Ascorbate
Urea cycle Ornithine Urea cycle L-Citrulline Urea cycle N-(L-Arginino)succinate Urea cycle Urate
Polyamines Putrescine Polyamines Spermidine Polyamines Spermine
Aminosugars N-Acetylneuraminate Arginine and proline metabolism N-Acetylornithine Arginine and proline metabolism Phosphocreatine Arginine and proline metabolism Creatine Arginine and proline metabolism trans-4-Hydroxy-L-proline Arginine and proline metabolism N-Succinyl-L-citrulline
Sulfur metabolism Taurine Sulfur metabolism Hypotaurine Sulfur metabolism Cys-Gly Sulfur metabolism 3-Sulfino-L-alanine Sulfur metabolism L-Methionine S-oxide
Figure 28 Global metabolomic analysis of CD147 and MCT4 knockdown PANC1 cells
94
Figure 28 Global metabolomic analysis of CD147 and MCT4 knockdown PANC1 cellsdCluster analysis of normalized MS results for indicated cell lines. shMCT4 cells were treated with DMSO or 20 µM MCT1 inhibitor for 72 hrs. Cells were lysed and data were collected as described in methods. shCD147 data were collected same as for the other samples but in a separate experiment. All intensity values were normalized to their respective shCTRL samples run alongside the indicated experiments. GENE cluster analysis was performed on the normalized values.
95
Figure 29 Schematic representation of cross-linking/MS approach used for the identification of possible CD147 interacting partners Wild type PANC1 cells were used for the analysis and two independent biological replicates were analyzed for each condition.
96
Figure 30 Cross-linking/MS identifies several distinct families of possible CD147 interacting partners (A) WBs conformation of the specificity of pull-downs. Samples were prepared as described in methods and the ability to pull down CD147 was confirmed via Western blotting. Input lines provided loading controls. (B) Gene ontology (GO) terms analysis for all 153 proteins identified as potential CD147 interacting partners in cross- linking/MS experiments. Calculation of over-represented GO terms was performed using the entire human genome database as a background (threshold count = 2, EASE score = 0.1). Only terms with a p value < 0.05 are listed.
97
Figure 31 The conformation of the interaction between CD147-ECD and CD98hc-ECD is inconclusive (A) WB with CD98hc specific antibody on PANC1 cells transfected with CD147-GFP and GFP treated with DSS for 30 min or DMSO, following cell lysis and pull- down with GFP-nanobody beads. Proteins were eluted with SDS loading buffer and processed for WBs. (B) 1H/15N HSQC NMR titration experiment for the uniformly 15N labeled CD147-ECD and unlabeled CD98hc ectodomain (aa 212-630, CD98hc-ECD) Increased concentrations of unlabeled CD98hc-ECD were titrated into 15N labeled protein (black - free protein).
98
Figure 32 Conformation of CD147 interactions via Western blotting PANC1 cells were transfected with CD147-GFP and GFP, treated with DSS for 30 min or DMSO followed by cell lysis and pull-down with GFP-nanobody beads. Proteins were eluted with SDS loading buffer and processed for WBs. Membranes were incubated with indicated primary antibodies (A-F, MCT4, MCT1, PMCA1, ICAM1, ASCT2 and LAT1) followed by corresponding secondary antibodies. Arrows indicate the location of the correct molecular weight band for each identified proteins. Cross-linked fractions are also marked.
99
Figure 33 PLA confirms interactions between CD147 and proteins identified in cross- linking/MS experiment (A) in situ proximity ligation assay (PLA) confirms interactions (< 40nm) of CD147 with target proteins, indicated as red dots in shCTRL cells and the reduction of such interactions when CD147 is depleted in shCD147 cells. Cell nuclei were counterstained with Hoechst. Scale bar 10µm. (B) PANC1 shCTRL and different shCD147 cells were screened for indicated protein expressions via Western blotting. β-actin provided loading control and only a representative β-actin blot is shown. (C) Immunofluorescence staining of shCTRL and shCD147 #1 PANC1 cells for the expression of indicated proteins. Arrows point to membrane localization. Localization of cell membranes was determined from phase-contrast microscopy. Cell nuclei were counterstained with Hoechst. Scale bar 10µm.
100
Figure 34 Proteasomal inhibitor treatments protect CD147 interacting partners from degradation
101
Figure 34 Proteasomal inhibitor treatments protect CD147 interacting partners from degradation (A) Western blot analysis for indicated proteins in PANC1 shCTRL or shCD147 cells treated with DMSO (0 µM MG132) or increasing concentrations of MG132 (proteasomal inhibitor). (B) Cells were treated with DMSO or 10 µM MG132 for 6hrs following IF staining with antibodies specific for the identified proteins. Cell nuclei were counterstained with Hoechst. Scale bar is 10 µm
102
Figure 35 Restoration of CD147 expression rescues the interacting proteins expression and activity
103
Figure 35 Restoration of CD147 expression rescues the interacting proteins expression and activity Indicated cell lines were stably transduced with lentiviral particles encoding CTRL (empty vector), wild type CD147 or glycosylation incapable mutant (CD147NG) (A) CD147 re-expression was detected by Western blotting. β-actin provided loading control. (B) Cell growth was assessed by counting live cells two days post seeding. Bars are ± SEM, n=3. (C) IF staining with CD147 specific antibody. Bottom panel shown wild type PANC1 cells transfected with GFP or CD147 fusion constructs. Cell nuclei were counterstained with Hoechst. Scale bar is 10 µm. (D) Western blots with antibodies specific for the indicated proteins. WBs marked with a star are from the same membranes, performed at the same time, but the images were cropped to remove two middle lines. β- actin provided loading control. (E) Metabolomics analysis for intracellular lactate levels. See Materials and Methods for experimental details. Bars are ± SEM, n=3, ***p<0.001.
104
Figure 36 CD147 depletion prompts PANC1 cells to reprogram glucose and glutamine metabolism
105
Figure 36 CD147 depletion prompts PANC1 cells to reprogram glucose and glutamine 13 13 15 metabolism Metabolite labeling experiments with C6-glucose and C6, N2- glutamine in PANC1 shCTRL, shCD147 and shMCT4 cells (A-D). See Materials and Methods for experimental details. Proteins detected in SILAC experiments are in green. G6P - glucose 6- phosphate, G3P - glyceraldehyde 3-phosphate, GL6P - 6-phospho 6-gluconolactone, 6PG - 6-phosphogluconate, PP - pentose phosphates, S7P - sedoheptulose 7-phosphate, αKG - α-ketoglutarate, PPP - pentose phosphate pathway, TCA cycle -tricarboxylic acid cycle, GLUT1 - glucose transporter 1, HKDC1 - hexokinase containing domain 1, PKM2 - pyruvate kinase isoform 2, G6PD - glucose 6-phosphate dehydrogenase. Bars are ± SEM, n=3, *p<0.05, **p<0.01 ***p<0.001.
106
Figure 37 The loss of CD147 expression leads to several phenotypic changes associated with cell growth inhibition (A) Cell morphology and size was evaluated using automated cell counter in control and shCD147 cells. Scale bar is 150 µm. (B) Indicated PANC1 cells were fixed and stained with Propodium Iodine (Millipore) and analyzed by fluorescence activated cell sorting (FACS). (C) Cells were analyzed by FACS after Annexin V staining (apoptotic marker). (D) Immunofluorescence staining of control and CD147 knockdown cells. Cell nuclei were counterstained with Hoechst. Scale bar is 10 µm.
107
Figure 38 PANC1 shCD147 cells utilize glycolytic intermediates to generate cell building blocks
108
Figure 38 PANC1 shCD147 cells utilize glycolytic intermediates to generate cell 13 13 15 building blocksMetabolite labeling experiments with C6-glucose and C6, N2- glutamine in PANC1 shCTRL, shCD147 and shMCT4 cells (A-B). See Materials and Methods for experimental details. Proteins detected in SILAC experiments are in green. αKG - α- ketoglutarate, PPP - pentose phosphate pathway, TCA cycle -tricarboxylic acid cycle, PYRC - pyrroline-5-carboxylase synthetase, PHGDGH - phosphoglycerate dehydrogenase. Bars are ± SEM, n=3, *p<0.05, **p<0.01 ***p<0.001.
109
Figure 39 The summary of the consequences of CD147 depletion on mesenchymal PDAC cells In CD147 expressing cells several CD147 interacting partners are protected from proteasomal degradation and trans-located to the cellular membrane. The absence of CD147 leads to the degradation of its interacting partners and reversal of the EMT. NA - nucleotides, AA - amino acids, ER - Endoplasmic reticulum.
110
CHAPTER III
A NEW MECHANISM OF CYCLPHILIN A INTERNALIZATION BY CANCER CELLS
Introduction and review of the literature
Cyclophilin A
Cyclophilin A (CypA) belongs to a family of highly conserved proteins (cyclophilins) that comprises 18 human isoforms120. Cyclophilins are ubiquitously expressed in all prokaryotic and eukaryotic cells. Most cyclophilins, including CypA are peptidyl-prolyl cis- trans isomerases (PPIase) that catalyze a reversible cis-trans isomerization of prolines
(Figure 40). CypA is an 18 kDa protein composed of an eight-stranded antiparallel β-barrel with two α helices surrounding the barrel from each site. The active site of CypA includes a conserved arginine residue (Arg55) and a mutation of this residue to alanine leads to approximately 90% inhibition of CypA PPIase activity167. The immunosuppressant drug cyclosporine A (CsA) binds to CypA with high affinity and irreversibly inhibits its activity. The binding of CsA to CypA is followed by formation of a ternary complex with calcineurin and subsequent inhibition of gene regulation and its PPIase activity120.
CypA is found in most cellular compartments, especially in the cytosol where it comprises ~0.1-0.6% of total cytosolic protein pool168. Perhaps the most well known function of CypA is its ability to assist proteins during folding process and, hence, its chaperone function168, 169. Additionally, CypA has been implicated in the regulation of cardiovascular diseases, cancer progression, viral replication of human immunodeficiency virus 1 (HIV-1), hepatitis C and B, and influenza A170-173. CypA is highly expressed in PDAC and intracellular targets of CypA have been proposed to include external regulated kinase (ERK)1/2174,
Janus-activated kinase 2175, and interleukin-2 tyrosine kinase176. Increased expression of
CypA leads to chemotherapy resistance in multiple cancers177, illustrating the importance of therapeutically targeting CypA during cancer progression. Furthermore, CypA deficient mice
111
are resistant to cyclosporine induced immunosuppresion, exhibit reduction in atherosclerosis, decrease in oxidative stress and matrix degradation178.
Initially characterized as an intracellular protein, CypA has also been detected in the extracellular regions due to the secretion from cells in response to inflammatory stimuli and cell death168. Indeed, elevated levels of extracellular CypA (herein referred as exCypA) have been identified in a variety of inflammatory diseases, including: sepsis and rheumatoid arthritis179. CypA, unlike other cyclophilins, does not contain a signal peptide sequence leading to its secretion via a non-canonical pathway i.e. vesicle shedding180. ExCypA has been proposed to induce autocrine/paracrine pathways via activating communication with its intracellular form, however the specifics of this activity have not been characterized. Studies show that exCypA sustains a high chemotactic activity in regard to leukocytes and macrophages stimulation and this activity might be linked to its autocrine/paracrine function181, 182. It has been proposed that the activity of exCypA is mediated via an interaction with a single pass transmembrane glycoprotein, CD147 (extensively reviewed in chapter II). This interaction has been proposed to be within the extracellular part of CD147, in particular with the residues in direct proximity of the membrane (Pro211). Interestingly, the binding affinity is in a low milimolar range, and thus, such weak interaction might not support direct signaling between exCypA and CD14753, 120. Certainly, this low affinity binding supports the possibility of the existence of other factors that mediate the exCypA/CD147 interaction or another exCypA receptor. Indeed, our group has previously showed that in some cells, stimulation of CD147 depleted cells with exCypA leads to the same level of signaling as in the control cells183.
The specifics of exCypa activity are not well characterized. In particular, exCypA has been shown to interact with cell surface heparans, but the contribution of this interaction to its signaling remains elusive183, 184. Saphire et al. illustrated that a CypA/heparan interaction is necessary for HIV infection but the contribution of heparan derived oligosaccharides to
112
other exCypA signaling pathways has not been depicted184. One possibility could be that the exCypA interaction with CD147 is mediated by the presence of glycans. This is especially plausible since CD147 exhibits high level of glycosylation within its extracellular region and the previously reported binding affinity, obtained from studies with unglycoslyated CD147, is relatively weak. Another question that persists is how does exCypA signal to activate its intracellular signaling pathways and is CD147 required for its activity? Li et al, previously showed that exCypA can stimulate proliferation and ERK 1/2 signaling in PDAC cell lines via
CD147 dependent pathway, however, the specifics of this activity have not been fully characterized185. In addition, our group revealed that exCypA is important for NFκB activity and cytokine secretion in those cells and that exCypA mediated ERK 1/2 signaling was
PDAC cell type specific183. Importantly, in each case the molecular determinants of exCypA activity in regard to pancreatic cancer were not fully depicted. A growing spectrum of evidence implicates intracellular CypA in endocytic pathway mechanism via its involvement in viral infection or toxin internalization, however exCypA signaling in regard to endocytic pathways and its relation to PDAC progression has not been well charaterized184, 186.
Endocytosis
Enodyctosis is a form of transport, in which cells internalize membrane bound or extracellular cargo via an active transport mechanism. As opposed to passive transport, active transport requires energy and, thus, endocytosis is an energy dependent process.
Recent studies have implicated endocytosis as one of the emerging pathways important in the control of cellular homeostasis and proliferation187. Many signaling processes are governed by the deregulation of endocytic pathways, typically due to receptor internalization mechanisms. Once inside the cells, receptors are sorted into different compartments for degradation or recycling. This sorting process is important in determining the fate of the implicated signaling cascade. Ligand-activated internalization of receptor tyrosine kinases
113
(RTKs) and G-protein coupled receptors (GPCRs) have been extensively studied with many examples of endocytic regulation. One such example is EGFR signaling, where receptor internalization upon ligand (EGF) binding attenuates EGFR mediated signaling events188.
Other transduction pathways influenced by endocytosis include Wnt and Notch transduction networks189.
Thus far, several different types of internalization pathways have been described, which can be characterized into two major groups, clathrin mediated endocytosis (CME) and clathrin independent endocytosis (CIE)190. The main type of endocytosis, CME, involves clathrin coated pits that are formed upon ligand activated receptor binding on the cell surface. Once signal is received, clathrin coated vesicles invaginate, pinch off, and internalization initiates. Subsequently, fused vesicles disassemble their clathrin coat and become endosomes to later fuse or develop into other vesicles, i.e. late endosomes, lysosomes. Although always activated by a receptor-ligand response, CME can be utilized by cells for activating signaling pathways, hormone internalization or virus infection190.
Several CIE process have been described with some being independent of receptor interactions. In a caveolin dependent processes, lipid rafts cluster on the plasma membrane to form caveolae buds, which then carry the cargo through cell membrane via receptor dependent or independent process. Macropinocytosis and phagocytosis are examples of receptor independent processes. Macropinocytosis is a process of internalization of large areas of plasma membrane to form small pits containing extracellular fluids and small molecules191. Phagocytosis, which encompasses engulfing of large cellular debris and pathogens, is a process specific for mammalian immune cells including neutrophils, monocytes and macrophages192. Recent discoveries propose varying functions for different endocytic pathways in regulating disease progression although, receptor mediated processes are commonly studied in regard to signal transduction pathways, macropinocytosis has been lately linked to providing building blocks for cells193. This
114
process is associated with internalization of nutrients important for metabolic processes, especially critical for fueling cancer cell growth and proliferation.
Study rationale
Recent years have witnessed an elevated detection of endocytic pathways and their involvement in disease progression. Additionally, it is increasingly evident that endocytosis has many effects on cellular pathways and, equally, that signaling via transmembrane receptors and other factors regulate the endocytic processes. Numerous extracellular proteins have been investigated for their role in receptor-mediated endocytosis, however this possibility has not been explored for extracellular CypA. Interestingly, exCypA signaling is linked to a membrane bound receptor, CD147 and CD147 has been proposed to interact with caveolin, a critical player in CIE. The existence of exCypA-CD147-caveolin connection could suggest an involvement of exCypA in endocytic signaling and, hence, propose a novel model of this extracellular protein activity. Thus, we aimed to test the possibility of endocytic signaling in exCypA function and the contribution of CD147 to this process. We also assessed the specific mechanism of such pathway and the functional consequences of this signaling in regard to PDAC progression.
Results
Cyclophilin A is internalized by pancreatic cancer cells and the internalization process is cell type specific
A multitude of functions have been proposed for exCypA, in regard to autocrine signaling and activation of several intracellular pathways important for disease progression.
Here, we investigated the mechanism of exCypA signaling in relation to possible endocytic signaling. exCypA was fused to the N-terminus of GFP and this fusion protein was expressed and purified from prokaryotic cells. We then stimulated different pancreatic cancer cell lines (PANC1, BxPC3, L3.6pl and L3.5pl) with increasing concentrations of
115
exCypA-GFP. As can be seen in Figure 41A we detected an internalization of CypA-GFP but not GFP alone. Furthermore, this internalization process was cell type specific (Figure
41B). From four different PDAC cell types tested (PANC1, data not shown), we detected intracellular CypA-GFP in two of the cell lines. Interestingly, L3.6pl and L3.5 cells are both derived from the same parental cell line and thus, carry very similar characteristics194. In particular, L3.6pl and L3.5 are both fast growing epithelial cells that were detected to exhibit metastasis from pancreas to liver and spleen, respectively. In contrast, BxPC3 and PANC1 are less metastatic with longer tumor latency rate136. This difference in the specific characteristics of tested cells, could potentially explain the cell type specificity of this internalization process.
We next tested the dose dependency of this process in order to obtain further information of a possible receptor that exCypA could be engaging for cellular uptake. An activation of a receptor on a cell surface would suggest that with increasing concentrations of exCypA receptor binding sites would ultimately become saturated and internalization of
CypA-GFP would reach equilibrium. Indeed, with increasing concentrations of CypA-GFP we observed an increase in internalization with an eventual saturation of the measured signal at higher concentrations (Figure 41C). This is consistent with an idea that exCypA requires a receptor for entering the cells. Since receptor mediated endocytosis is an energy involving process, we proceeded to test the internalization process by performing the experiment at lower temperature (4 oC) where energy supply is significantly decreased.
L3.6pl cells treated with CypA-GFP at 37 oC engulfed significant amount of exCypA, while cells treated in cold suppressed the internalization process with most signal detected on the cell membrane (Figure 41D, green dots). This appearance of signal on the membrane, in cold temperature treated cells, points to the possibility that exCypA binds to a receptor on the cell surface independently of the temperature, but cannot enter the cells due to the inability to activate the endocytic pathway.
116
Lastly, we wanted to determine what is the fate of exCypA once it enters the cells.
This is an important aspect of exCypA studies that aim to establish the functional consequences of exCypA signaling and internalization. After treatment with CypA-GFP, we stained cells with markers for different organelles and evaluated the co-localization pattern between CypA-GFP and different organelles. Figure 41E illustrates that exCypA co-localizes with early endosomes and the trans-Golgi compartment. This cellular localization pattern would suggest that after entering the cells, CypA-GFP does not directly undergo proteasomal degradation, which takes place in the lysosome, but rather is retained by the cells for possible other functions.
Clathrin mediated endocytosis is the key entry pathway for exCypA
Our dose dependency and temperature treatments data already indicates that exCypA is internalized by the cells via receptor mediated process. Thus, we next wanted to characterize the specific pathway utilized by cells to uptake CypA-GFP. To do this, we inhibited different endocytic pathways and monitored internalization process by immunofluorescence. We used genestien and intracellular potassium depletion to inhibit caveolin and clathrin dependent processes, respectively. Genistein is a tyrosine kinase inhibitor that specifically inhibits caveolae-mediated processes195. Clathrin interaction with its adaptor proteins is potassium dependent, and thus, depletion of intracellular potassium by incubation of cells in hypertonic media prevents clathrin coated pits formation. As depicted in Figure 42A, the uptake of CypAGFP was unaffected in genestein treated cells, but CypA-GFP uptake was significantly diminished in cells exposed to intracellular potassium depletion. Our analysis revealed that CypA-GFP uptake was attenuated upon
CME inhibition but unaffected in the other conditions (Figure 42A). We then proceeded to test the colocalization of internalized CypA-GFP with clathrin. We stained CypA-GFP treated cells with clathrin antibody and evaluated cells for co-localization. Based on our evaluation,
117
we observed possible co-localization events (Figure 42B). However, considering the large amount of staining for clathrin, only a very small percentage of signal showed co-localization patterns. This is not completely surprising, since clathrin disassembles very rapidly after vesicles pinch off of the membrane and, thus, the observed staining might correspond to intracellular clathrin transport vesicles between Golgi and other intracellular organelles.
To fully confirm the clathrin-mediated mechanism for exCypA uptake, we performed an acute knockdown of clathrin with a pool of different siRNAs in PDAC cells lines. Again, we focused our experiments to L3.6pl and L3.5 cells, since these cell lines internalize exCypA. Western blotting confirmed a 50% depletion of clathrin in these cell lines (Figure
42C). We then treated clathrin-depleted (shCHC) and control (shCTRL) cells with CypA-
GFP or GFP and monitored internalization by IF after cell fixation. As it can be seen in
Figure 42C, there was a substantial diminishment in internalized CypA-GFP upon clathrin depletion. Interestingly, we detected an increase in membrane bound CypA-GFP in shCHC cells. Likewise, the low temperature experiment (Figure 41D), it is plausible that exCypA is binding to its receptor on the cell surface, but due to endocytosis inhibition by clathrin downregulation, it is not being taken up by the cells.
ExCypA internalization process is CD147 dependent
To test the contribution of the only known CypA receptor, CD147 to the internalization mechanism, we utilized previously generated L3.6pl CD147 knockdown cells
(Figure 21). We first confirmed the downregulation of CD147 in those cells (Figure 43A) followed by treatment with CypA-GFP or GFP. FACS analysis revealed striking diminishment of exCypA uptake consistent with the level of CD147 depletion (Figure 43A).
Because CypA interaction with CD147 has been shown to take place through the extracellular region, we also treated cells with anti-CD147 antibody or anti-IgG control followed by CypA-GFP/GFP treatment. Figure 43C illustrates that this treatment lead to
118
decreased uptake of exCypA exemplifying the importance of the CD147 extracellular region to this interaction. We followed up this discovery with a biophysical analysis were we tested the interaction between CypA and the extracellular glycosylated CD147. Previous studies, from our group and others, indicated that CypA weakly interacts with extracellular membrane proximal proline (Pro211) of CD147, but this interaction might be enhanced by the presence of CD147 glycans99. This is even more likely since CypA has also been shown to interact with hyparan-derived glycans183. To test this interaction we utilized previously generated unlabeled glycosylated CD147 (Figure 18) and uniformly labeled 15N CypA.
Unfortunately, NMR titration experiments revealed no direct interaction between CypA and glycosylated CD147 (Figure 43C). This lack of detected interaction may be explained by the absence of Pro211 in our CD147-ECD construct, but also indicates that CypA does not directly interact with the CD147 glycans as we had expected.
The functional consequences of exCypA internalization are inconclusive
To establish the functional contribution of exCypA internalization, we tested the activity of exCypA in regard to PDAC cell lines. Because CD147 mediated signaling is correlated with an alteration of MMP secretion in multiple cancers, we tested the possible contribution of exCypA to MMP cellular release90, 91, 196. Although, we observed an increase in MMP1 secretion upon exCypA treatment in L3.6pl, we also detected a significant decrease in MMP1 extracellular levels in L3.5 cells (Figure 44A). Another aspect of exCypA activity is its influence on cell proliferation and chemotaxis. Thus, we measured proliferation rate and chemotaxis after treatment with varying doses of exCypA in different PDAC cell lines. We did not observe any significant difference in proliferation rates upon exposure to exCypA (Figure 44B). Additionally, chemotactic activity of exCypA in regard to L3.6pl cells was only modestly increased, but again this was not the case for other cells that internalize exCypA (Figure 44C). Lastly, to test the possibility that exCypA treatment leads to a
119
decrease in signaling due to receptor (CD147) internalization, as suggested for EGFR and its ligand, EGF. We incubated L3.6pl cells with exCypA and measured extracellular CD147 expression at different time points post treatment. We did not observe any significant change in CD147 extracellular expression as indicated in Figure 44D, suggesting that
CD147 is not internalized along with exCypA or the internalization/recycling process is too fast for the measurement. Thus, at this stage, we have not been able to obtain the functional consequences of exCypA update in PDAC cells.
Discussion
ExCypA has been implicated in the regulation of several cellular processes however, the specific molecular pathway of its activity has not been depicted. Here we identified a new mechanism of exCypA function, in regard to the utilization of an endocytic pathway by pancreatic cancer cell lines. Our analysis revealed that PDAC cells are capable of internalizing exCypA via clathirn-mediated endocytosis. This was a surprising and intriguing discovery, as there has been no evidence for internalization previously reported. CME is a vital process implicated in the regulation of a variety of receptor-mediated signaling cascades. ExCypA cellular uptake via CME pathway suggests that exCypA mediated signaling is facilitated by a specific receptor, which likely plays a role in the regulation of its activity. Indeed, CD147 downregulation or blockage with an antibody was able to suppress exCypA internalization event. Although, we were not able to confirm a direct interaction between CypA and CD147 and the involvement or glycans in this interaction, we showed that exCypA cellular uptake is dependent on CD147 expression. This is an important observation that can further aid in discovering the specific pathways altered by exCypa endocytosis.
We also presented evidence that pancreatic cancer cells internalize exCypA in a cell type specific manner. This finding suggests a particular cellular selectivity for exCypA
120
uptake in PDAC cell lines. In our hands, two cell lines (L3.6pl and L3.5), derived from the same source, were capable of taking up exCypA, while the other two tested cell lines
(BxPC3 and PANC1) did not. Interestingly, L3.6pl and L3.5 have already been shown to exhibit high level of cytokine and MMP secretion, and thus, a possibility for an inhibitory action of exCypA in regard to decreased cytokine/MMP production in those cells, would be an attractive justification for the observed phenotypes183. Though, our analysis did not provide evidence for such inhibitory pathway existence, but rather presented conflicting results in determining the functional consequences of exCypA activity. Taken together, our study identified a novel molecular feature of exCypA in pancreatic cancer. The discoveries presented here may shed light on some of the signaling pathways affected by exCypA and open new doors in terms of discovering novel roles for exCypA activity in PDAC progression.
Future directions
We have provided the first evidence for clathrin mediated endocytic uptake of exCypA in pancreatic cancer cell lines. However, we were unable to identify the functional consequences of this internalization process. An attractive idea that could be tested next is the possibility of an inhibitory pathway upon exCypA uptake. The cells that exhibit internalization process (L3.6pl and L3.5) also produce large amounts of MMPs, cytokines and growth factors. Therefore, studies aimed at identifying such inhibition events could be undertaken. Alternatively, a new study has suggested that intracellular CypA modulates Crk phosphorylation and cell migration197. Thus, exCypA internalization may culminate in Crk phosphorylation inside the cell in order to regulated cellular migration. In addition, it would also be important to identify whether this internalization process takes place in other cancers where CD147 and CypA overexpression has also been reported.
121
Figure 40 Prolyl isomerization reaction catalyzed by PPIase enzymes
122
Figure 41 CypA is internalized by pancreatic cancer cell lines (A) L3.6pl cells were treated with indicated concentrations of GFP or CypA-GFP for 4 hrs at 37oC. Internalization was measured by fluorescence activated cell sorting (FACS) and signal is represented as normalized mean fluorescence intensity (MFI). Bars are ± SEM, n=3, **p<0.01,***p<0.001. (B) Internalization was measured in the indicated cells after treatment (30 min at 37oC) via FACS. Signal was normalized and is represented as fold change over GFP treated cells. Bars are ± SEM, n=3, **p<0.01,***p<0.001. (C) Dose response of GFP or CypA-GFP internalization in L3.5 cells measured using FACS. Mean fluorescence intensity (MFI) was normalized to PBS treatment. Values are ± SEM, n=4, **p<0.01,***p<0.001. (D) L3.6pl cells were treated with 5 uM GFP or CypA-GFP for 30 min at the indicated temperatures. Cells were fixed and internalization visualized by IF. Nuclei were stained with Hoechst. Scale bar is 5 µm (E) Post CypA-GFP treatment cells were fixed, permeabalized and stained with organelle specific markers: Golgi - WGA-FITC, early endosome - EEA1-Cy5 and late endosome/lysosome - Lamp1-Cy5. Nuclei were stained with Hoechst. Scale bar is 5 µm
123
Figure 42 CypA is internalized via clathrin mediated endocytosis (A) L3.6pl cells were treated with 10 µM genstein to inhibit clathrin independent endocytosis (CIE) or exposed to intracellular K+ depletion to inhibit clathrin mediated endocytosis (CME) before treatment with 5 µM CypA-GFP. Post treatment cells were fixed and imaged to visualize internalized CypA- GFP. Nuclei were stained with Hoechst. Scale bar is 5 µm. (B) Magnified image of cells treated with CypA-GFP and after fixing stained with clathirn antibody. Arrows point to possible co-localization. (C) L3.5 cells were transfected with siRNA encoding scramble control or clathrin (CHC) specific sequence. Clathrin downregulation was monitored via WB with clathrin specific antibody. β-actin provided loading control. (D) Same cells as in C were treated with 5 µM CypA-GFP and stained with clathirn antibody after fixation. Membrane was stained with Alexa-568 conjugated wheat germ agglutinin (WGA). Nuclei were stained with Hoechst. Scale bar is 5 µm.
124
Figure 43 Internalization process depends on CD147 expression but does not rely on CD147 ectododomain (A) L3.6pl were stably transfected with shRNAs encoding scramble control or two CD147 shRNAs. CD147 depletion was confirmed with CD147 specific antibody. β-actin provided loading control. (B) Same cells as in A were treated for 30 mins with 5 µM CypA-GFP or GFP and internalization was confirmed by FACS. Mean fluorescence intensity (MFI) was normalized to GFP treatment. (C) L3.6pl cells were pretreated with anti-CD147 or anti-IgG antibodies and following 30 min treatment with 5 µM CypA-GFP or GFP internalization was visualized by IF. WGA was used for membrane staining. Scale bar is 5 µm. D. 1H/15N HSQC NMR titration experiment for the uniformly 15N labeled CypA and unlabeled glycosylated CD147 ectodomain (aa 22-205, CD147-ECDgly) Increased concentrations of unlabeled CD147-ECDgly were titrated into 15N labeled protein (black - free protein).
125
Figure 44 The functional consequences of exCypA internalization are inconclusive (A) Indicated cells were stimulated with increasing concentrations of exCypA or buffer control for 24hrs in serum free media. Following treatment, conditioned media was collected and tested for MMP1 secretion via ELISA assay. Representative results are shown. Bars are ± SEM, n=4. (B) L3.5 cell proliferation was measured after 24hrs treatment of cells with indicated concentrations of exCypA or buffer control in serum free or serum containing media. Bars are ± SEM, n=8. (C) Boyden chamber assay was used to measure cell migration towards serum free media containing the indicated concentrations of exCypA. Bars are ± SEM, n=4. (D) Receptor internalization was measured after incubating cells with 5 uM exCypA for the indicated time points. Cells were incubated with anti-CD147 antibody in cold and signal detected by FACS. Values are ± SEM, n=4.
126
CHAPTER IV
MATERIALS AND METHODS
Reagents and plasmids
All antibodies and their corresponding applications are listed in Table 6. MCT1 inhibitor, ARC-155858 (TOCRIS, UK), MG132 (Fisher Scientific) and Genetsin (Sigma) were all dissolved in DMSO and used at the indicated concentrations. For each compound dissolved in DMSO the corresponding concentration of DMSO was used as a control never exceeding 0.5% DMSO.
pLenti CMV/TO Neo empty (w215-1) was a gift from Eric Campeau (Addgene plasmid # 17485). CD147-GFP plasmid was cloned in house into pcDNA3.1-GFP with an N- terminal signal peptide and a C-terminal GFP. Point mutations and deletion constructs were generated with Quick Change Site directed mutagenesis kit (Agilent Technologies, Santa
Clara, CA) according to manufacture’s description.
Protein expression and purification
CXCL8 constructs
DNA encoding the 72 amino acid wild-type human CXCL8 was commercially purchased (Genewiz, South Planfield, NJ, USA), PCR amplified, and subsequently ligated into the pET15b plasmid between NdeI and BamH1 sites for expression in E. Coli strain
BL21 (DE3). The pET15b plasmid expression vector (Novagen Inc., Madison, WI) containing a 6xHis tag and thrombin cleavage site was used for initial CXCL8 protein expressions to generate CXCL8Synthetic, containing a post cleavage GSHM overhang. In addition, a Factor Xa cleavage site (IEGR) was engineered directly C-terminal to the thrombin cleavage site (LVPRGS) in order to obtain the wild-type (no overhang) fully active sequence of CXCL8 (starting with SAK). Two CXCL8 constructs were inserted into these vectors, which included wild-type human full length CXCL8 and a monomeric variant
127
(CXCL8M) obtained by performing site directed mutagenesis to acquire L25Y and V27R double mutant24. CXCL8 protein (wild-type and monomeric) was grown in LB or M9 minimal media supplemented with 15N-ammonium chloride and/or 13C-glucose for generation of unlabeled, 15N or 15N/13C labeled samples. Proteins were refolded from the insoluble fractions as previously described37 and subsequently purified via a Ni-affinity column followed by cleavage of the 6xHis tag with thrombin or Factor Xa (NEB, Ipswich, MA) and then size exclusion chromatography. All purifications were conducted on an AKTA FPLC system (GE Healthcare). Final samples for NMR contained 0.2-1.0 mM protein in NMR buffers: phosphate (50 mM Na3PO4, 150 mM NaCl pH 6.5) or HEPES (50 mM HEPES, 50 mM NaCl, pH 7.0) supplemented with 5% D2O. hCXCR1 peptide
For peptide purification, a previously described protocol was applied with slight modifications99. Briefly, standard PCR amplification methods were used to generate a construct containing amino acids corresponding to residues 9-29 of hCXCR1
(MWDFDDLNFTGMPPADEDYSP). The hCXCR1 peptide construct was cloned into a pet15b vector containing the small GB1 protein derived from streptococcal protein G, a
6xHis tag followed by a thrombin cleavage site, and then residues 9-29 of hCXCR1. This fusion protein was expressed in BL21 (DE3) cells in LB or isotope enriched media (see above for media composition). After Ni-affinity, thrombin was used to release the peptide and followed by HPLC purification. Peptide fractions from HPLC purification were lyophilized, re-suspended in NMR buffer (phosphate or HEPES), and the pH adjusted appropriately. The identity of peptide was also confirmed via mass spectrometry using a
MALDI-TOF instrument.
128
CD147-ECD purification
CD147-ECD was expressed as previously described92, 99. Briefly, CD147-ECD (aa
22-205) was ligated into pet15b and expressed in E. Coli strain BL21 (DE3). Cells were lysed via sonication in 35 ml of 100 mM Tris, NaCl and 1 mM EDTA, pH 7.5, centrifuged at
12,000g and insoluble fraction was collected. The insoluble fraction was further sonicated in
5 M guanidine, 100 mM Tris, 100 mM NaCl, 2 mM 2-mercaptoethanol, pH 7.5, centrifuged again and dialyzed into refolding buffer (1 M arginine, 100 mM Tris, 100 mM NaCl, pH 7.5) for 48 hrs. Samples were subsequently dialyzed into 100 mM Tris, 300 mM NaCl, pH 7.5 for additional 12 hrs for subsequent dialysis into 50 mM Na2HPO4, pH 7.5, 500 mM NaCl, 10 mM imidazole. Soluble fraction was then applied to Ni affinity column and eluted with 50 mM
Na2HPO4, pH 7.5, 500 mM NaCl, 400 mM imidazole. Elutions comprising 6xHis-tagged
CD147-ECD were then concentrated to approximately 2-3 ml, the thrombin tag was cleaved off with 3 units of thrombin incubated overnight at room temperature, and the protein was applied to a superose-75 sizing column equilibrated in NMR buffer (50 mM Na2HPO4, pH
6.5) or 1X PBS. Elutions comprising CD1247-ECD were concentrated to approximately 0.5-
1.5 mM for subsequent analysis.
CD98hc purification
CD98hc ORF was purchased from Functional Genomics Facility, UCD Cancer
Center. The extracellular region of CD98hc (aa 212-630) was cloned in house into pet15b containing an N-terminal 6xHis and thrombin cleave site. CD98hc-ECD in pet15b was expressed in E. Coli strain BL21 (DE3). Cell pellets were lysed via sonication in nickel buffer
(50 mM Na2HPO4, pH 7.5, 500 mM NaCl, 10 mM imidazole), applied to a 10-20 ml Ni- sepharose column, and eluted with nickel elution buffer (50 mM Na2HPO4, pH 7.5, 500 mM
NaCl, 400 mM imidazole). Elutions comprising 6xHis-tagged CD98hc were then concentrated to approximately 2-3 ml, the thrombin tag was cleaved off with 3 units of
129
thrombin incubated overnight at room temperature, and the protein was applied to a superose-75 sizing column equilibrated in NMR buffer (50 mM Na2HPO4, pH 6.5). Elutions comprising CD98hc were concentrated to approximately 0.5-1.5 mM with 95% H2O/5% D2O for subsequent NMR analysis.
GFP-nanobody purification and conjugation to beads
6xHis tagged GFP-nanobody was synthesized into pJ401 plasmid (DNA2.0, Menlo
Park, CA). GFP-nanobody in pJ401 was expressed in E.coli strain BL21 (DE3) and purified via Ni-affinity column and ion exclusion chromatograph in 50 mM Na2HPO4, 150mM NaCl, pH 6.5. Purified GFP-nanobody was conjugated to NHS-activated Sepharose Fast Flow resin (GE Healthcare, Pittsburgh, CA) according to manufactures protocol. Briefly, resin was washed once with 1M HCl and twice with 0.2 M NaHCO3, 500mM NaCl, pH 8.4. 1 mg of
GFP-nanobody was then added to 0.5 ml of resin and incubated over-night in 4 oC with gentle agitation. Next day any unreacted groups were blocked with 0.1 M Tris, pH 8.4 and resin was washed and stored in 20% ethanol for further use.
CypA purification
CypA was expressed and purified as previously published.198, 199 Briefly, cells were lysed in 25 mM MES, pH 6.1, with 1 mM DTT, purified over an SP Sepharose exchange column, dialyzed into 50 mM Tris, 1 mM EDTA pH 6.8 and flowed through a Q Sepharose column, followed by size exclusion chromatography in 1x PBS buffer
CypA-GFP and GFP purifications
DNA encoding the wild-type human CypA was cloned into pet15b plasmid, which contained a N-terminal, 6xHis tag followed by thrombin cleavage site and C-terminal GFP.
CypA-GFP or GFP proteins were grown in LB media and proteins were purified via Ni-
130
affinity followed by 6xHis release with thrombin. Final purification step included size exclusion chromatography in buffer containing 50 mM Tris, 500 mM NaCl, pH 7.5.
Mammalian CD147-ECD expression and purification
For high yield mammalian expression system pCEP4 vector was utilized (Thermo
Fisher, Grand Island, NY). For full length CD147 construct (CD147-FL), residues corresponding to a signal peptide (amino acids 1-21 of human CD147) followed by FLAG tag, 6xHis affinity tag, a TEV cleavage site, and amino acids 22-269 of human CD147, were cloned into a mammalian expression vector (pCEP4) between NdeI and BamH1 sites using
InFusion cloning strategy (Clontech). CD147 ectodomain construct (CD147-ECD) was generated by an insertion of a stop codon after amino acids 205 in CD147FL pCEP4 construct to obtain Ig1-Ig2 (aa 22-205) construct.
The FreeStyle 293-F cells (Invitrogen) were sub-cultured in FreeStyle 293 expression medium (Invitrogen) in a shaker under standard humidified conditions (37oC and
5% CO2). Transient transfections with PEI (Polyethylenimine, Fisher Scientific) were performed as previously described with slight modifications200. Briefly, the day before transfection cells were passaged into fresh media to final concentration of 1x10^6 cells/ml and allowed to grow overnight. On the day of transfection cells were spun down and re- suspended to 2.5-3x10^6 cells/ml in fresh medium supplemented with 3ug/ml of DNA and
9ug/ml of PEI, and allowed to grow overnight. Next day, cells were diluted to 1x10^6 cells/ml with fresh medium supplemented with VPA (Valproic acid, Sigma) at a final concentration of
2.2 mM. Cells were then allowed to grow for 5 days and conditioned media was collected on day 5.
Secreted proteins were purified from culture media after removal of cells via centrifugation for 20min at 8000rpm and culture media filtration through 0.22µm. Typically,
1L of total culture media was purified at a time. Immediately before IMAC chromatography
131
0.2L of 5X Ni buffer A1 (250 mM phosphate, 2.5 M NaCl, 50 mM Imidazole, pH 7.5) was added to the media and the whole volume was applied to pre-equilibrated 10ml Ni-high affinity column using AKTA FPLC system (GE Healthcare). 6xHis tagged protein was eluted with increasing concentration of Ni buffer A1 + 400mM Imidazole and concentrated to about
1ml for subsequent gel filtration chromatography (S75 column). Gel filtration was performed in final buffer containing 50 mM phosphate 50 mM NaCl pH 6.5 for NMR or 1X PBS for cell culture treatments Protein purity was evaluated by SDS-page gel electrophoresis and measurement of OD ratio at 260/280.
NMR spectroscopy
Assignments and standard NMR experiments
All NMR spectra were collected at 25 °C on a Varian 600 MHz or 900-MHz spectrometer with samples supplemented with 5% D2O. Samples utilized for assignment purposes contained 0.5–1 mM protein, while all samples used for titrations contained 0.2-
0.5 mM protein with the indicated final concentration of the titrant. All spectra were processed using NMRPipe software201 and analyzed using CCPNmr software202. Unless otherwise noted, all pulse sequences were obtained from standard Varian Biopack libraries.
Standard multidimensional NMR experiments HNCACB, CBCA(CO)NH and HNCA were collected for each construct used, and both a 3D-15N-NOESY and a 3D-15N-TOCSY (total correlated spectroscopy) were used to confirm each spin system. Backbone sequential assignments included a CBCA(CO)NH, HNCACB and HNCO. For CXCL8 and CXCL8M 3D
15N-edited and 13C-edited NOESY-HSQC experiments were also acquired (NOESY experiments employed a tmix = 70 ms and tmix= 50 ms, respectively). For assignments of
CXCL8/CXCL8M bound to peptide, peaks were followed during titration experiments.
132
Relaxation experiments
For relaxation experiments, standard R1 and R2 relaxation experiments were applied with recycle delays of 2.5 s at either 900 or 600 MHz. Relaxation delays for R1 experiments were 0.01, 0.1, 0.3, 0.5, 0.7, 0.9, and 1.1 s, and relaxation delays for R2 experiments were
0.01, 0.03, 0.05, 0.07, 0.11 and 0.13 s. While compensating pulses prior to the recycle delay were utilized for R2 measurements to account for potential sample heating, all relaxation experiments were arrayed within a single experiment to also account for any potential field inhomogeneities. Correlation times for each amide and free or bound protein were calculated using R2/R1 relaxation rate ratios as described in Larsson et al.31. In-house scripts that combined both NMRPipe and Xmgrace software were used to fit and visualize, respectively, all relaxation rates using peak heights. The previously published correlation times (Table 2) were plotted versus molecular weight (MW) and the experimental MWs were calculated by fitting the experimental τc values into a least-squares linear regression
15 equation (τc = 0.59*MW). N-R2-CPMG pulse sequences were applied for relaxation dispersion experiments with the delay time of 0.05 s for CXCL8 and 0.07 for CXCL8M and
H(S/M)QC pulse sequences were applied to measure exchange-induced shifts on a Varian
600 MHz spectrometer collected at 25oC21, 203. Relaxation compensated pulse sequences for all 15N-R2-CPMG experiments were also employed.
Molecular dynamics simulations
Replica-averaged restrained molecular dynamics simulations44, 45, 47, 204 were performed using GROMACS205 coupled with PLUMED206 and Almost207. The simulations were carried out using the Amber99SB*-ILDN force field208 and the TIP3P water model. A time step of 2 fs was used together with LINCS constraints209. The van der Waals and electrostatic interactions were cut-off at 0.9 nm, and long-range electrostatic effects were treated with the particle mesh Ewald method210. All the simulations were done in the
133
canonical ensemble by keeping the volume fixed and by thermosetting the system with the
Bussi thermostat211. The starting conformations for both the monomer and the dimer were taken from an X-ray structure212 (PDB code 1IL8). In the case of the monomer the L25Y and
V27R mutations were modeled with PyMol213. The structures were protonated and solvated with 7000 water molecules in a dodecahedron box of 244 nm3 of volume. The energy of the system was first minimized and then the temperature was increased to 300 K in two separate steps, in the first one a 50 ps long simulation was performed by keeping fixed the heavy atoms of the protein, and successively a second 200 ps long simulation was performed without restraints. The density of the system has been relaxed by a 200 ps long run using the Berendsen barostat. The molecular dynamics simulations were carried out with replica-averaged chemical shift restraints, using four replicas47. The starting structures for the four replicas were selected as the final structure from four 1ns simulations. Each replica was evolved through a series of annealing cycles between 300 K and 400 K, each cycle being composed of 100 ps at 300 K, 100 ps of linear increase in the temperature up to
400 K, 100 ps of constant temperature molecular dynamics simulations at 400 K and 300 ps of linear decrease in the temperature to 300 K. Only structures from the 300 K portions of the simulations were taken into account for analysis. Each replica was evolved for 150 ns.
The resulting ensemble is composed by all the structures sampled at 300 K by all the replicas after discarding the first 10 ns.
Mammalian cell culture
Cell lines and culture conditions
Human HeLa and HEK293FT cells were purchased from University of Colorado
Cancer Center Core and were cultured in DMEM and RPMI media, respectively, supplemented with 10% FBS and 0.1g/ml penicilin/streptomyocin. Normal immortalized pancreatic cells (NT1, NT2, NT5) and pancreatic cancer cell lines (HS-766T BxPC3,
134
PANC1, MiaPaCa2, L3.6pl and L3.5) were a generous gift from Dr. Colin Weekes. Cells were cultured in DMEM media supplemented with 10% FBS, 0.1 g/ml penicilin/streptomyocin and 1% non-essential amino acids (Thermo Fisher). All cells were routinely tested for mycoplasma contamination.
SILAC cell culture conditions
For SILAC analysis DMEM was supplemented with dialyzed serum and heavy - Arg-
13C/Lys-13C, or light - Arg-12C/ Lys-12C (ThermoScientific). shCD147 cells were cultured in the heavy media and shCTRL cells were cultured in the light media. Cells were allowed 8 doublings before 98% isotope incorporation was achieved. The incorporation was tested via
MS and for final analysis cells were mixed 1:1 based on the total protein content.
Stable cell lines generations
Lentiviral particles encoding shRNA directed against BSG or SLC16A3 genes
(MISSION™ TRC shRNA, Sigma) were used to transduce L.6pl, PANC1 and MiaPaCa2 cells using standard protocols. Stable integration was achieved by puromycin selection
(2mg/ml). Knockdown was confirmed by western blot analysis and qRT-PCR, indicating approximately 70% - 90% depletion. Similarly, lentiviral particles encoding human BSG
ORF in pLenti CMV/TO vector or empty vector were transduced into knockdown cells using standard protocols. Stable integration was achieved by neomocin selection (300 µg/mL) and cell sorting with CD147 targeting antibodies. Gene expression was confirmed by
Western blot analysis and qRT-PCR indicating 30-50% rescue.
Transfections
All transfections were performed using Lipofectamine 3000 (Invitrogen) as per manufacturer protocol. For siRNA transfections, control siRNA or a pool of siRNA targeting clathirn heavy chain (siCHC, MISSION™ siRNA, Sigma) was transfected into indicated cells
135
using Lipofectamine 3000 (Invitrogen). Cells were assayed 48hrs post transfection by
Western blotting for clathrin downregulation (50%).
Immunofluorescence
Cells were grown in 8-well chamber slides and following treatment or transfections, when applicable, cells were fixed in either cold 100% methanol for 8mins in -20oC or 2% paraformaldeyde in 1X PBS for 15mins in RT. Cells were then washed with 1X PBS, and blocked with Knudsen Buffer (1x PBS, 0.5%BSA, 0.5% NP-40, 1mM MgCl2, 1mM NaN3).
Cells were stained with the specific primary antibodies following appropriate secondary antibodies incubations. DNA was visualized using Hoechst 33342 (Sigma). Coverslips were mounted using Citifluor AF-1 (Ted Pella) mounting media.
Intracellular organelle staining
Cells were grown in 8-well chamber slides or collagen coated coverslips (24-well plates) and following treatment, cells were fixed in cold methanol for 8mins in -20oC. For intracellular staining cells were permeabilized with 0.1 % Triton and blocked with 1x PBS,
0.5%BSA. For Golgi visualization, cells were stained with Alexa-568-WGA for 15 mins in
RT. Membrane staining was done with Alexa-568-WGA as described above, but without the permeabilization step. For endosome or lysosome visualization cells were stained with anti-
EEA1 or anti-LAMP1, respectively, following appropriate secondary antibodies incubations.
DNA was visualized using Hoechst 33342 (Sigma). Coverslips were mounted using Citifluor
AF-1 (Ted Pella) mounting media.
Proximity ligation assay (PLA) analyses
Protein interactions were analyzed in situ, using Duolink II PLA detection kit (Sigma
Aldrich, St. Louis, MO) as per manufacturers instructions. Briefly, cells were processes same as for immunofluorescence, but Duolink specific secondary antibodies were utilized
136
instead following probe ligation and amplification. The antibodies used to test the interactions are listed in Table 6. The specificity of each interaction was tested by omitting one of the primary antibodies in the assay preparation.
Imaging and image analysis
All images were visualized with a Nikon Ti Eclipse inverted microscope (Nikon,
Melville, NY) with a Nikon 100x PlanApo numerical aperture 1.4 objective. Images were captured with and Android iXon electron-multiplying charge-coupled device (CCD) 888E camera (Andor Technologies, Belfast, United Kingdom). Image analysis was performed using NIS Elements imaging software (Nikon) and Image J.
Annexin-V staining
Cells were grown to 80–90% confluence prior to FACS analysis. The cells were trypsinized, washed twice with FACS buffer (1X PBS containing 1% BSA and 5 mM EDTA), and resuspended in the same buffer to a concentration of 10 x 105 cells/100 µL. Direct- labeled antibody, Annexin-V-FITC (Thermo Fisher) was added to samples at a dilution of
1:50 and incubated in the dark for 30 min on ice. Subsequently, cells were washed twice with FACS buffer and resuspended in FACS buffer containing 1x PI to stain nuclei, followed by flow cytometry using a Guava EasyCyte flow cytometer (EMD Millipore, Billerica, MA).
Side scatter and forward scatter profiles were used to eliminate cell doublets. FITC labeled
IgG isotope control was used as a negative control.
Cell cycle arrest analysis
24 hrs before the analysis cells were plated in 6-well plates. On the day of analysis cells were trypsinized, washed twice with FACS buffer, permeabilized with 2% paraformaldehyde and stained with PI. Cells were then fixed with 1% PFA/1X PBS and
137
analyzed by flow cytometry using a Guava EasyCyte flow cytometer (EMD Millipore,
Billerica, MA). Side scatter and forward scatter profiles were used to eliminate cell doublets.
Cell growth and cell size analysis
On day 1, cells were plated into 6-well plates at 0.2 x 106 cells/well in complete media and allowed to adhere for 24hrs. On day 2, and every day after for next 3-4 days, cells were trypsinized with 100 µL trypsin, re-suspended in 900 µL of complete media and seeded back into the same plate wells. Each day a 10 µL aliquot was stained with Tryptan blue and cell count, and size was monitored using Countess cell counter (Thermo Fisher).
All cell growth data was normalized to day 1.
Cell migration
Migration was assayed using 8-µm pore Boyden chambers (Costar, Boston, MA).
200 uL of cells, at 1 x 106 cells/ml was plated in the top chamber of the transwell insert and
500 uL of media containing the indicated chemoatractant (CypA, FBS) was placed in the lower chamber. Cells were allowed to migrate for 4hrs at 37oC in a humid atmosphere at 5%
CO2. After incubations, cells were removed from top well and filters were incubated in 250 uL of 1X Cell dissociation Solution (Trevigen, MA). Filters were discarded and dissociated cells were lysed by freeze-thawing. Finally, cell lysates were incubated with CyQuant reagent (Thermo Fisher) and signal was read at 480/520 using fluorescence plate reader.
Enzyme-linked immunosorbent assay (ELISA)
Cells were plated in 12-well plates and allowed to adhere for 24 hrs. Cells were then stimulated (in serum free media) with the indicated concentrations of the specified stimulants (exCypA, CD147-ECD, CD147-ECDgly) for 24 hrs. Conditioned media was the collected and 100 µL was applied to the ELISA plates. Measurements were carried out as per manufacturer's protocol (ELISA Tech, Aurora, CO).
138
Cytokine profiling
Conditioned media cytokine and MMP levels were assayed by chemiluminescence- based sandwich immunoassay according to manufacturer's instructions (MesoScale
Diagnostics). Plates were analyzed on a Sector 2400 Imager (MesoScale Diagnostics).
Western blot
Cell lysis and Western blotting was performed as previously described214. Briefly, for total cell lysates, subconfluent cells were harvested directly into RIPA (0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 50 mM Tris, pH 7.5) buffer and lysed by freeze-thawing. Protein content was quantified using BCA protein assay reagent (Thermoscientific) and 20-30 µg of extract subjected to SDS/PAGE to assay for the indicated proteins. All antiboides used for WBs are listed in Table 6.
Immunoprecipitations (IPs)
For immunoprecipitations, equal amount of total cells lysate (500 µg) was applied to
40 ul of GFP-nanobody conjugated beads and incubated on a rotating shaker for 1 hr in 4 oC. Beads were then washed four times with 500 ul of RIPA buffer and 40 ul of 1X SDS loading buffer was added to the beads, heated to 100 oC for 10 min and resolved on
NuPAGE Bis-Tris 4−12% gradient gel (Invitrogen, CA). Gels were stained with Comassie for
Proteomics analysis (see Proteomics analysis) or transferred to PVDF membranes and processed for Western blotting as previously described 214.
Proteasomal inhibitor treatments
Cells were grown to 60–70% confluence prior to treatments. The cells were treated with indicated concentrations of MG132 in full serum media for 6 hrs at 37oC in a humid atmosphere at 5% CO2.
139
MCT1 inhibitor treatments
Cells were grown to 40–60% confluence prior to treatments. The cells were treated with indicated concentrations of ARC-155858 (MCT1inh) in full serum media for 6 hrs at
o 37 C in a humid atmosphere at 5% CO2. Following treatment cells were processed for microscopy or lysed for WBs.
CypA-GFP and GFP treatments
Cells were grown to 60–70% confluence prior to treatments. For microscopy, cells were grown on collagen-coated coverslips and for FACS analysis cells were grown in 96- well plates. The cells were treated with CypA-GFP or GFP in full serum media immediately prior to analysis. Buffer treatment (50 mM Tris, 500 mM NaCl, pH 7.5) was used as a control.
FACS analysis of CypA-GFP or GFP treated cells
Post treatments cells were washed three times wth 1X PBS, trypsinized, washed twice more with PBS with and resuspended in 1X PBS containing 1x PI to stain nuclei. Flow cytometry was performed using a Guava EasyCyte flow cytometer (EMD Millipore, Billerica,
MA). Side scatter and forward scatter profiles were used to eliminate cell doublets.
Cold temperature internalization experiments
Cells were grown to 60–70% confluence prior to treatments. On the day of treatment plates were moved to 4 oC and allowed to equilibrate for 1 hrs. CypA-GFP or GFP was added to cold full serum media and cells were treated for additional hour in the 4 oC.
Following treatment cells were processed for microscopy.
140
Potassium depletion
Potassium depletion was performed as previously described215. Briefly, cells were split into 24-well plates and allowed to adhere for 24 hrs. Next day, cells were washed with
+ K depletion buffer (140 mM NaCl, 20 mM HEPES, 1 mM CaCl2, 1mM MgCL2, 1g/L D- glucose, pH 7.5) and incubated with 50% K+ depletion buffer/ddH2O for 15 mins in 37oC.
Control cells were incubated in the same buffer supplemented with 10 mM KCl. Following the incubations, cells were treated with CypA or CypA-GFP in the indicated buffers and processed for microscopy.
Receptor internalization
On the day of analysis the cells were trypsinized, washed twice with complete media, and resuspended in the media to a concentration of 10 x 105 cells/100 µL. Cells were incubated with 5uM CypA-GFP for 5,10,15,20 and 30 mins and incubations were stopped by centrifugation at 4oC and washes with ice cold 1X PBS and stored in ice cold FACS buffer containing 1x PI to stain nuclei. Cells were stored on ice for subsequent FACS analysis.
Cells incubated with GFP were used for background signal and background signal was subtracted form all readouts. FACS was performed using a Guava EasyCyte flow cytometer
(EMD Millipore, Billerica, MA). Side scatter and forward scatter profiles were used to eliminate cell doublets.
Global metabolomics cell culture conditions
Equal number of cells was plated into 6-well plates in 1 ml of complete media/well and allowed to adhere for 24 hrs. On the day of the experiment cells were trypsinized, counted and extracted with 1ml/2 x 106 cells of extraction buffer (methanol:acetonitrile:water
5:3:2) immediately prior to analysis, 20 µL of conditioned media was also collected and metabolites were extracted by addition of 480 µL of extraction buffer. Samples were then
141
agitated at in 4oC for 30 min following by centrifugation at 10,000 g for 10 min at 4oC. Protein and lipid pellets were discarded and metabolite fractions were stored at -80 oC for further analysis.
Metabolic tracing experiments
Equal number of cells was plated into 6-well plates and allowed to adhere for 24-48 hrs. Cells where then washed with 1X PBS and media replaced with media containing 10%
13 13 dialyzed serum and the indicated isotopically enriched carbon sources ( C6-glucose or C6,
15 N2- glutamine). The control experiment was performed alongside with cells exposed to
12 12 14 media containing 10% dialyzed serum and C6-glucose and C6, N2- glutamine. Glucose experiments were performed for 1 hr and glutamine for 24 hrs. After treatment, samples were processed same as for the global metabolomic analysis. 20 µL of conditioned media was also used for the analysis.
Mass spectrometry
Proteomics analysis
SDS-PAGE analysis Samples were loaded onto a 1.5 mm thick NuPAGE Bis-Tris 4−12% gradient gel
(Invitrogen). The BenchMark™ Protein Ladder (Invitrogen) was used as a protein molecular mass marker. The electrophoretic run was performing by using MES SDS running buffer, in an X-Cell II mini gel system (Invitrogen) at 200 V, 120 mA, 25 W per gel for 30 minutes. The gel was stained using SimplyBlue™ SafeStain (Invitrogen) stain and de-stained with water according to the manufacturer’s protocol.
In-gel tryptic digestion After excision, gel pieces were destained in 200 µL of 25 mM ammonium bicarbonate in 50 % v/v acetonitrile for 15 min and washed two times with 200 µL of 50%
(v/v) acetonitrile. Disulfide bonds in proteins were reduced by incubation in 10 mM
142
dithiothreitol (DTT) at 60 °C for 30 min and cysteine residues were alkylated with 20 mM iodoacetamide (IAA) in the dark at room temperature for 45 min. Gel pieces were subsequently washed with 100 µL of distilled water followed by addition of 100 µL of acetonitrile and dried on SpeedVac (Savant ThermoFisher). Then 100 ng of trypsin was added to each sample and allowed to rehydrate the gel plugs at 4 °C for 45 min and then incubated at 37 °C overnight. The tryptic mixtures were acidified with formic acid up to a final concentration of 1%. Peptides were extracted two times from the gel plugs using 1% formic acid in 50% acetonitrile. The collected extractions were pooled with the initial digestion supernatant and dried on SpeedVac (Savant ThermoFisher).
Mass Spectrometry A nanoflow HPLC instrument (Easy nLC 1000 UHPLC, Thermo Fisher Scientific) was coupled on-line to a Q Exactive mass spectrometer (Thermo Fisher Scientific) with a nanoelectrospray ion source (Proxeon). Peptides were separated on a self-made 10 cm C18 analytical column (100 µm x 10 cm) packed with 2.7 µm Phenomenex Cortecs C18 resin.
After equilibration with 3µL 5% acetonitrile 0.1% formic acid, the peptides were separated by a 50 min gradient from 2% to 35% acetonitrile with 0.1% formic acid at 350nL/min. LC mobile phase solvents and sample dilutions used 0.1% formic acid in water (Buffer A) and
0.1% formic acid in acetonitrile (Buffer B) (Chromasolv LC–MS grade; Sigma-Aldrich, St.
Louis, MO). Data acquisition was performed using the instrument supplied Xcalibur™
(version 3.0) software. The mass spectrometer was operated in the positive ion mode, in the data–dependent acquisition mode. In one scan cycle, peptide ions were first scanned by full
MS at resolution 70,000 (FWHM at m/z 200), and then the top 10 intensive ions were sequentially subjected to HCD fragmentation and detected at resolution 17,500.
143
Database searching, protein identification MS/MS spectra were extracted from raw data files and converted into mgf files using a PAVA script (UCSF, MSF, San Francisco, CA). These mgf files were then independently searched against mouse SwissProt database using an in-house Mascot™ server (Version
2.2.06, Matrix Science). Mass tolerances were +/- 10ppm for MS peaks, and +/- 0.1 Da for
MS/MS fragment ions. Trypsin specificity was used allowing for 1 missed cleavage. Met oxidation, proline hydroxylation, protein N-terminal acetylation, and peptide N-terminal pyroglutamic acid formation were allowed for variable modifications while carbamidomethyl of Cys was set as a fixed modification.
Scaffold (version 4.3.2, Proteome Software, Portland, OR, USA) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability as specified by the
Peptide Prophet algorithm. Protein identifications were accepted if they could be established at greater than 99.0% probability and contained at least two identified unique peptides.
Gene ontology analysis Gene ontology analysis was performed using the DAVID Bioinformatics Database
(DAVID Bioinformatics Resources, http://david.abcc.ncifcrf.gov/) with the human gene name identifiers of proteins identified to be significantly (p<0.05, protein identified in at least two out of three biological replicates) upregulated or downregulated SILAC analysis. Calculation of over-represented GO terms was performed using the entire list of identified proteins as background (threshold count = 2; EASE score = 0.1). Terms with a p-value <0.05 were selected, log10- transformed and hierarchically clustered using GENE software.
Metabolomics analysis
Metabolomics analysis was performed as previously described216. Breifly, 10 µl of samples were injected into an UHPLC system (Ultimate 3000, Thermo, San Jose, CA, USA)
144
and separated during a 3 minute isocratic gradient at 250 µl/min (mobile phase: 5% acetonitrile, 95% 18 mΩ H2O, 0.1% formic acid) in a Kinetex C18 column (150x1 mm i.d.,
1.7 µm particle size – Phenomenex, Torrance, CA, USA). The UHPLC system was coupled online with a QExactive system (Thermo, San Jose, CA, USA), scanning in Full MS mode (2
µscans) at 70,000 resolution in the 60-900 m/z range, 4kV spray voltage, 15 sheath gas and
5 auxiliary gas, operated either in negative and positive ion mode. Calibration was performed before each analysis against positive or negative ion mode calibration mixes
(Piercenet – Thermo Fisher, Rockford, IL, USA) to ensure sub ppm error on the intact mass.
Metabolite assignments were performed through the software Maven1 (Princeton, NJ, USA), upon conversion of .raw files into .mzXML format through MassMatrix (Cleveland, OH,
USA). The software allows for peak picking, feature detection and metabolite assignment using the KEGG pathway database. Assignments were further confirmed by chemical formula determination from isotopic patterns and accurate intact mass and retention times against a subset of standards including commercially available glycolytic and Krebs cycle intermediates, amino acids, glutathione homeostasis and nucleoside phosphates (SIGMA
Aldrich, St. Louis, MO, USA).
Statistical analysis
Binding isotherms calculations
GraphPad Prism software (GraphPad Prism Software Inc., la Jolla, CA) was used to determine the binding isotherms. Only residues exhibiting fast exchange were included with amides that exhibit chemical shift changes above the digital resolution of the indirect dimension (~0.6 ppm). These residues were simultaneously fit using nonlinear least square fit to the equation Δδobs = Δδsat x ([Ligand]tot + [Protein]tot + Kd – (([Ligand]tot + [Protein]tot +
2 0.5 Kd) – 4 x [Ligand]tot x [Protein] ) / (2 x [Protein]tot), where Δδobs (normalized chemical
! ! change as defined by (5��!!) + (��!"!) ) ) is the observed Δδ at the given Ligand
145
concentration, Δδsat is the Δδ at saturation. Protein is the specific CXCL8 construct used at a constant total concentration ([Protein]tot) for each titration experiment and ([Ligand]tot is
CXCR1pep or CXCL8synthetic (see SFig. 2) at a specific total concentration for the respective titration point.
Metabolomics analysis
Relative quantitation was performed by exporting integrated peak areas values into
Excel (Microsoft, Redmond, CA, USA) for statistical analysis including T-Test (significance threshold for p-values < 0.05) and partial least square discriminant analysis (PLS-eDA), calculated through the macro MultiBase (freely available at www.NumericalDynamics.com).
Hierarchical clustering analysis (HCA) was performed through the software GENE-E (Broad
Institute). XY graphs were plotted through GraphPad Prism.
All other data statistical analysis
GraphPad Prism software was used for statistical analysis. Data are expressed as mean values SEM. Data were analyzed with Student's t-test between two groups or analysis of variance (ANOVA) coupled with post-hoc Bonferroni test for multiple pairwise comparisons. Probability values of P < 0.05 were considered to be statistically significant.
146
Table 6 Antibodies used in the study
Target Supplier Catalog number Application CD147 R&D Systems MAB972 WB CD147-FITC Ancell 376-040 FACS CD147 SCBT sc-21746 IF, PLA CD147 SCBT sc-13976 IF, PLA MCT4 SCBT sc-50329 WB, IF, PLA MCT4 SCBT sc-376139 IF, PLA MCT1 SCBT sc-50324 WB, IF, PLA E-cadherin Cell Signaling 3195 WB E-cadherin Invitrogen 33-4000 IF β-catenin Cell Signaling 8480 WB β-catenin BD 610154 IF Vimentin DBHB AMF-17b WB, IF EGFR Cell Signaling 2239 WB, IF, PLA ICAM Cell Signaling 4915 WB, IF, PLA PMCA1 Invitrogen PA1-914 WB, IF, PLA ASCT2 SCBT sc-99002 WB, IF, PLA CD98hc Cell Signaling 13180 WB, IF, PLA LAT1 Cell Signaling 5347 WB, IF, PLA GLUT1 Abcam ab652 IF ITGβ4 Cell Signaling 4707 WB, IF PKM2 Cell Signaling 4053 WB, IF G6PD Cell Signaling 8866S WB GFP SCBT sc-9996 WB HKDC1 Cell Signaling sc-243031 IF LAMP-1 Millipore AB2971 IF EEA1 Millipore 07-1820 IF GLUT1 Novus NB110-39113 IF Clathrin BD 610499 WB, IF B-actin Cell signaling 8457 WB B-actin Sigma Aldrich A5441 WB anti-mouse Cell Signaling 7076 WB anti-rabbit Cell-Signaling 7074S WB anti-goat SCBT sc-2020 WB Alexa-488 anti-rabbit Invitrogen A11008 IF Alexa-647 anti-rabbit Invitrogen A11011 IF Alexa-647 anti-mouse Invitrogen A21235 IF Alexa-568 anti-mouse Invitrogen A20037 IF Alexa-488 WGA Invitrogen W11261 IF
147
REFERENCES
(1) Kendrick AA, Holliday MJ, Isern NG, Zhang F, Camilloni C, Huynh C, Vendruscolo M, Armstrong G, and Eisenmesser EZ The dynamics of interleukin-8 and its interaction with human CXC receptor I peptide. Protein science : a publication of the Protein Society 23: 464-480 (2014).
(2) Waugh DJJ, and Wilson C The Interleukin-8 Pathway in Cancer. Clinical Cancer Research 14: 6735-6741 (2008).
(3) Allen SJ, Crown SE, and Handel TM Chemokine: receptor structure, interactions, and antagonism. Annual review of immunology 25: 787-820 (2007).
(4) Baldwin ET, Weber IT, St Charles R, Xuan JC, Appella E, Yamada M, Matsushima K, Edwards BF, Clore GM, Gronenborn AM, and et al. Crystal structure of interleukin 8: symbiosis of NMR and crystallography. Proceedings of the National Academy of Sciences of the United States of America 88: 502-506 (1991).
(5) Clore GM, Driscoll PC, Wingfield PT, and Gronenborn AM Low resolution structure of interleukin-1 beta in solution derived from 1H-15N heteronuclear three-dimensional nuclear magnetic resonance spectroscopy. Journal of molecular biology 214: 811- 817 (1990).
(6) Yao S, Young IG, Norton RS, and Murphy JM Murine interleukin-3: structure, dynamics, and conformational heterogeneity in solution. Biochemistry 50: 2464-2477 (2011).
(7) Skelton NJ, Quan C, Reilly D, and Lowman H Structure of a CXC chemokine-receptor fragment in complex with interleukin-8. Structure with Folding & Design 7: 157-168 (1999).
(8) Clark-Lewis I, Schumacher C, Baggiolini M, and Moser B Structure-activity relationships of interleukin-8 determined using chemically synthesized analogs. Critical role of NH2-terminal residues and evidence for uncoupling of neutrophil chemotaxis, exocytosis, and receptor binding activities. The Journal of biological chemistry 266: 23128-23134 (1991).
(9) Baggiolini M, and Clark-Lewis I Interleukin-8, a chemotactic and inflammatory cytokine. FEBS letters 307: 97-101 (1992).
(10) Luppi F, Longo AM, de Boer WI, Rabe KF, and Hiemstra PS Interleukin-8 stimulates cell proliferation in non-small cell lung cancer through epidermal growth factor receptor transactivation. Lung Cancer 56: 25-33 (2007).
(11) Petreaca ML, Yao M, Liu Y, Defea K, and Martins-Green M Transactivation of vascular endothelial growth factor receptor-2 by interleukin-8 (IL-8/CXCL8) is required for IL- 8/CXCL8-induced endothelial permeability. Mol Biol Cell 18: 5014-5023 (2007).
(12) MacManus CF, Pettigrew J, Seaton A, Wilson C, Maxwell PJ, Berlingeri S, Purcell C, McGurk M, Johnston PG, and Waugh DJ Interleukin-8 signaling promotes translational regulation of cyclin D in androgen-independent prostate cancer cells. Mol Cancer Res 5: 737-748 (2007).
148
(13) Gilmore T, Gapuzan ME, Kalaitzidis D, and Starczynowski D Rel/NF-kappa B/I kappa B signal transduction in the generation and treatment of human cancer. Cancer Letters 181: 1-9 (2002).
(14) Holmes WE, Lee J, Kuang WJ, Rice GC, and Wood WI Structure and functional expression of a human interleukin-8 receptor. Science 253: 1278-1280 (1991).
(15) Rajarathnam K, Prado GN, Fernando H, Clark-Lewis I, and Navarro J Probing receptor binding activity of interleukin-8 dimer using a disulfide trap. Biochemistry 45: 7882- 7888 (2006).
(16) Park SH, Casagrande F, Cho L, Albrecht L, and Opella SJ Interactions of Interleukin-8 with the Human Chemokine Receptor CXCR1 in Phospholipid Bilayers by NMR Spectroscopy. J Mol Biol. 414: 194-203 (2011).
(17) Ravindran A, Joseph PR, and Rajarathnam K Structural basis for differential binding of the interleukin-8 monomer and dimer to the CXCR1 N-domain: role of coupled interactions and dynamics. Biochemistry 48: 8795-8805 (2009).
(18) Lipari G, and Szabo A Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 2. Analysis of experimental results. Journal of the American Chemical Society 104: 4559-4570 (1982).
(19) Carr HY, and Purcell EM Effects of Diffusion on Free Precession in Nuclear Magnetic Resonance Experiments. Physical Review 94: 630-638 (1954).
(20) Meiboom S, and Gill D Modified Spin‐Echo Method for Measuring Nuclear Relaxation Times. Review of Scientific Instruments 29: 688-691 (1958).
(21) Skrynnikov NR, Dahlquist FW, and Kay LE Reconstructing NMR spectra of "invisible" excited protein states using HSQC and HMQC experiments. Journal of the American Chemical Society 124: 12352-12360 (2002).
(22) Schlegel J, Armstrong GS, Redzic JS, Zhang F, and Eisenmesser EZ Characterizing and controlling the inherent dynamics of cyclophilin-A. Protein science : a publication of the Protein Society 18: 811-824 (2009).
(23) Clubb RT, Omichinski JG, Clore GM, and Gronenborn AM Mapping the Binding Surface of Interleukin-8 Complexed with an N-Terminal Fragment of the Type-1 Human Interleukin-8 Receptor. Febs Letters 338: 93-97 (1994).
(24) Daly TJ, LaRosa GJ, Dolich S, Maione TE, Cooper S, and Broxmeyer HE High activity suppression of myeloid progenitor proliferation by chimeric mutants of interleukin 8 and platelet factor 4. The Journal of biological chemistry 270: 23282-23292 (1995).
(25) Rajagopalan L, and Rajarathnam K Ligand selectivity and affinity of chemokine receptor CXCR1. Role of N-terminal domain. The Journal of biological chemistry 279: 30000- 30008 (2004).
149
(26) Leong SR, Kabakoff RC, and Hebert CA Complete mutagenesis of the extracellular domain of interleukin-8 (IL-8) type A receptor identifies charged residues mediating IL-8 binding and signal transduction. The Journal of biological chemistry 269: 19343- 19348 (1994).
(27) Fernando H, Chin C, Rosgen J, and Rajarathnam K Dimer dissociation is essential for interleukin-8 (IL-8) binding to CXCR1 receptor. Journal of Biological Chemistry 279: 36175-36178 (2004).
(28) Gerety SJ, Karpus WJ, Cubbon AR, Goswami RG, Rundell MK, Peterson JD, and Miller SD Class II-restricted T cell responses in Theiler's murine encephalomyelitis virus- induced demyelinating disease. V. Mapping of a dominant immunopathologic VP2 T cell epitope in susceptible SJL/J mice. J Immunol 152: 908-918 (1994).
(29) Burrows SD, Doyle ML, Murphy KP, Franklin SG, White JR, Brooks I, McNulty DE, Scott MO, Knutson JR, Porter D, and et al. Determination of the monomer-dimer equilibrium of interleukin-8 reveals it is a monomer at physiological concentrations. Biochemistry 33: 12741-12745 (1994).
(30) Fernando H, Nagle GT, and Rajarathnam K Thermodynamic characterization of interleukin-8 monomer binding to CXCR1 receptor N-terminal domain. Febs Journal 274: 241-251 (2007).
(31) Larsson G, Martinez G, Schleucher J, and Wijmenga SS Detection of nano-second internal motion and determination of overall tumbling times independent of the time scale of internal motion in proteins from NMR relaxation data. Journal of biomolecular NMR 27: 291-312 (2003).
(32) Lee D, Hilty C, Wider G, and Wuthrich K Effective rotational correlation times of proteins from NMR relaxation interference. J Magn Reson 178: 72-76 (2006).
(33) Rossi P, Swapna GV, Huang YJ, Aramini JM, Anklin C, Conover K, Hamilton K, Xiao R, Acton TB, Ertekin A, Everett JK, and Montelione GT A microscale protein NMR sample screening pipeline. Journal of biomolecular NMR 46: 11-22 (2010).
(34) Taylor GM, Ma L, Vogt VM, and Post CB NMR relaxation studies of an RNA-binding segment of the rous sarcoma virus gag polyprotein in free and bound states: a model for autoinhibition of assembly. Biochemistry 49: 4006-4017 (2010).
(35) Kay LE, Torchia DA, and Bax A Backbone dynamics of proteins as studied by 15N inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease. Biochemistry 28: 8972-8979 (1989).
(36) Kojima C, Ono A, Kainosho M, and James TL Quantitative measurement of transverse and longitudinal cross-correlation between 13C-1H dipolar interaction and 13C chemical shift anisotropy: application to a 13C-labeled DNA duplex. J Magn Reson 136: 169-175 (1999).
(37) Eisenmesser EZ, Millet O, Labeikovsky W, Korzhnev DM, Wolf-Watz M, Bosco DA, Skalicky JJ, Kay LE, and Kern D Intrinsic dynamics of an enzyme underlies catalysis. Nature 438: 117-121 (2005).
150
(38) Henzler-Wildman K, and Kern D Dynamic personalities of proteins. Nature 450: 964- 972 (2007).
(39) Boehr DD, Dyson HJ, and Wright PE An NMR perspective on enzyme dynamics. Chemical reviews 106: 3055-3079 (2006).
(40) Mulder FA, Skrynnikov NR, Hon B, Dahlquist FW, and Kay LE Measurement of slow (micros-ms) time scale dynamics in protein side chains by (15)N relaxation dispersion NMR spectroscopy: application to Asn and Gln residues in a cavity mutant of T4 lysozyme. Journal of the American Chemical Society 123: 967-975 (2001).
(41) Berjanskii MV, and Wishart DS A simple method to predict protein flexibility using secondary chemical shifts. Journal of the American Chemical Society 127: 14970- 14971 (2005).
(42) Cavalli A, Salvatella X, Dobson CM, and Vendruscolo M Protein structure determination from NMR chemical shifts. Proceedings of the National Academy of Sciences of the United States of America 104: 9615-9620 (2007).
(43) Shen Y, Lange O, Delaglio F, Rossi P, Aramini JM, Liu GH, Eletsky A, Wu YB, Singarapu KK, Lemak A, Ignatchenko A, Arrowsmith CH, Szyperski T, Montelione GT, Baker D, and Bax A Consistent blind protein structure generation from NMR chemical shift data. Proceedings of the National Academy of Sciences of the United States of America 105: 4685-4690 (2008).
(44) Camilloni C, Robustelli P, De Simone A, Cavalli A, and Vendruscolo M Characterization of the conformational equilibrium between the two major substates of RNase A using NMR chemical shifts. Journal of the American Chemical Society 134: 3968-3971 (2012).
(45) Camilloni C, De Simone A, Vranken WF, and Vendruscolo M Determination of secondary structure populations in disordered states of proteins using nuclear magnetic resonance chemical shifts. Biochemistry 51: 2224-2231 (2012).
(46) Xue Y, Ward JM, Yuwen T, Podkorytov IS, and Skrynnikov NR Microsecond time-scale conformational exchange in proteins: using long molecular dynamics trajectory to simulate NMR relaxation dispersion data. Journal of the American Chemical Society 134: 2555-2562 (2012).
(47) Camilloni C, Cavalli A, and Vendruscolo M Assessment of the use of NMR chemical shifts as replica-averaged structural restraints in molecular dynamics simulations to characterize the dynamics of proteins. The journal of physical chemistry. B 117: 1838-1843 (2013).
(48) Williams MA, Cave CM, Quaid G, Robinson C, Daly TJ, Witt D, Lentsch AB, and Solomkin JS Interleukin 8 dimerization as a mechanism for regulation of neutrophil adherence-dependent oxidant production. Shock 23: 371-376 (2005).
(49) Horcher M, Rot A, Aschauer H, and Besemer J IL-8 derivatives with a reduced potential to form homodimers are fully active in vitro and in vivo. Cytokine 10: 1-12 (1998).
151
(50) Das ST, Rajagopalan L, Guerrero-Plata A, Sai J, Richmond A, Garofalo RP, and Rajarathnam K Monomeric and dimeric CXCL8 are both essential for in vivo neutrophil recruitment. PloS one 5: e11754 (2010).
(51) Osborne MJ, and Wright PE Anisotropic rotational diffusion in model-free analysis for a ternary DHFR complex. Journal of biomolecular NMR 19: 209-230 (2001).
(52) Cole R, and Loria JP Evidence for flexibility in the function of ribonuclease A. Biochemistry 41: 6072-6081 (2002).
(53) Schlegel J, Redzic JS, Porter CC, Yurchenko V, Bukrinsky M, Labeikovsky W, Armstrong GS, Zhang F, Isern NG, DeGregori J, Hodges R, and Eisenmesser EZ Solution characterization of the extracellular region of CD147 and its interaction with its enzyme ligand cyclophilin A. Journal of molecular biology 391: 518-535 (2009).
(54) Markus MA, Dayie KT, Matsudaira P, and Wagner G Local mobility within villin 14T probed via heteronuclear relaxation measurements and a reduced spectral density mapping. Biochemistry 35: 1722-1732 (1996).
(55) Redzic JS, Armstrong GS, Isern NG, Jones DNM, Kieft SK, and Eisenmesser EZ The retinal specific CD147 Ig0 domain: from molecular structure to biological activity. J Mol Biol. 411: 68-82 (2011).
(56) Nucci NV, Marques BS, Bedard S, Dogan J, Gledhill JM, Jr., Moorman VR, Peterson RW, Valentine KG, Wand AL, and Wand AJ Optimization of NMR spectroscopy of encapsulated proteins dissolved in low viscosity fluids. Journal of biomolecular NMR 50: 421-430 (2011).
(57) Neuzillet C, Tijeras-Raballand A, Bourget P, Cros J, Couvelard A, Sauvanet A, Vullierme MP, Tournigand C, and Hammel P State of the art and future directions of pancreatic ductal adenocarcinoma therapy. Pharmacol Ther 155: 80-104 (2015).
(58) Hezel AF, Kimmelman AC, Stanger BZ, Bardeesy N, and Depinho RA Genetics and biology of pancreatic ductal adenocarcinoma. Genes Dev 20: 1218-1249 (2006).
(59) Wood LD, and Hruban RH Pathology and molecular genetics of pancreatic neoplasms. Cancer J 18: 492-501 (2012).
(60) Wolfgang CL, Herman JM, Laheru DA, Klein AP, Erdek MA, Fishman EK, and Hruban RH Recent progress in pancreatic cancer. CA Cancer J Clin 63: 318-348 (2013).
(61) Raimondi S, Lowenfels AB, Morselli-Labate AM, Maisonneuve P, and Pezzilli R Pancreatic cancer in chronic pancreatitis; aetiology, incidence, and early detection. Best Pract Res Clin Gastroenterol 24: 349-358 (2010).
(62) Moffitt RA, Marayati R, Flate EL, Volmar KE, Loeza SG, Hoadley KA, Rashid NU, Williams LA, Eaton SC, Chung AH, Smyla JK, Anderson JM, Kim HJ, Bentrem DJ, Talamonti MS, Iacobuzio-Donahue CA, Hollingsworth MA, and Yeh JJ Virtual microdissection identifies distinct tumor- and stroma-specific subtypes of pancreatic ductal adenocarcinoma. Nat Genet 47: 1168-1178 (2015).
152
(63) Pan JJ, and Yang MH The role of epithelial-mesenchymal transition in pancreatic cancer. J Gastrointest Oncol 2: 151-156 (2011).
(64) Haynes J, Srivastava J, Madson N, Wittmann T, and Barber DL Dynamic actin remodeling during epithelial-mesenchymal transition depends on increased moesin expression. Mol Biol Cell 22: 4750-4764 (2011).
(65) van Roy F, and Berx G The cell-cell adhesion molecule E-cadherin. Cell Mol Life Sci 65: 3756-3788 (2008).
(66) Suzuki M, Shigematsu H, Nakajima T, Kubo R, Motohashi S, Sekine Y, Shibuya K, Iizasa T, Hiroshima K, Nakatani Y, Gazdar AF, and Fujisawa T Synchronous alterations of Wnt and epidermal growth factor receptor signaling pathways through aberrant methylation and mutation in non small cell lung cancer. Clinical cancer research : an official journal of the American Association for Cancer Research 13: 6087-6092 (2007).
(67) Lee CH, Hung HW, Hung PH, and Shieh YS Epidermal growth factor receptor regulates beta-catenin location, stability, and transcriptional activity in oral cancer. Mol Cancer 9: 64 (2010).
(68) Schroeder JA, Adriance MC, McConnell EJ, Thompson MC, Pockaj B, and Gendler SJ ErbB-beta-catenin complexes are associated with human infiltrating ductal breast and murine mammary tumor virus (MMTV)-Wnt-1 and MMTV-c-Neu transgenic carcinomas. The Journal of biological chemistry 277: 22692-22698 (2002).
(69) Barriere G, Fici P, Gallerani G, Fabbri F, and Rigaud M Epithelial Mesenchymal Transition: a double-edged sword. Clin Transl Med 4: 14 (2015).
(70) Jones S, Zhang X, Parsons DW, Lin JC, Leary RJ, Angenendt P, Mankoo P, Carter H, Kamiyama H, Jimeno A, Hong SM, Fu B, Lin MT, Calhoun ES, Kamiyama M, Walter K, Nikolskaya T, Nikolsky Y, Hartigan J, Smith DR, Hidalgo M, Leach SD, Klein AP, Jaffee EM, Goggins M, Maitra A, Iacobuzio-Donahue C, Eshleman JR, Kern SE, Hruban RH, Karchin R, Papadopoulos N, Parmigiani G, Vogelstein B, Velculescu VE, and Kinzler KW Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science 321: 1801-1806 (2008).
(71) Eser S, Schnieke A, Schneider G, and Saur D Oncogenic KRAS signalling in pancreatic cancer. British journal of cancer 111: 817-822 (2014).
(72) Chang ZG, Wei JM, Qin CF, Hao K, Tian XD, Xie K, Xie XH, and Yang YM Suppression of the epidermal growth factor receptor inhibits epithelial-mesenchymal transition in human pancreatic cancer PANC-1 cells. Dig Dis Sci 57: 1181-1189 (2012).
(73) Heerboth S, Housman G, Leary M, Longacre M, Byler S, Lapinska K, Willbanks A, and Sarkar S EMT and tumor metastasis. Clin Transl Med 4: 6 (2015).
(74) Karamitopoulou E Role of epithelial-mesenchymal transition in pancreatic ductal adenocarcinoma: is tumor budding the missing link? Front Oncol 3: 221 (2013).
153
(75) Ying H, Kimmelman AC, Lyssiotis CA, Hua S, Chu GC, Fletcher-Sananikone E, Locasale JW, Son J, Zhang H, Coloff JL, Yan H, Wang W, Chen S, Viale A, Zheng H, Paik JH, Lim C, Guimaraes AR, Martin ES, Chang J, Hezel AF, Perry SR, Hu J, Gan B, Xiao Y, Asara JM, Weissleder R, Wang YA, Chin L, Cantley LC, and DePinho RA Oncogenic Kras maintains pancreatic tumors through regulation of anabolic glucose metabolism. Cell 149: 656-670 (2012).
(76) Cohen R, Neuzillet C, Tijeras-Raballand A, Faivre S, de Gramont A, and Raymond E Targeting cancer cell metabolism in pancreatic adenocarcinoma. Oncotarget 6: 16832-16847 (2015).
(77) Warburg O On the Origin of Cancer Cells. Science 123: 309-314 (1956).
(78) Gatenby RA, and Gillies RJ Why do cancers have high aerobic glycolysis? Nat Rev Cancer 4: 891-899 (2004).
(79) Dang CV Links between metabolism and cancer. Genes Dev 26: 877-890 (2012).
(80) Metallo CM, Gameiro PA, Bell EL, Mattaini KR, Yang J, Hiller K, Jewell CM, Johnson ZR, Irvine DJ, Guarente L, Kelleher JK, Vander Heiden MG, Iliopoulos O, and Stephanopoulos G Reductive glutamine metabolism by IDH1 mediates lipogenesis under hypoxia. Nature 481: 380-384 (2012).
(81) Stincone A, Prigione A, Cramer T, Wamelink MM, Campbell K, Cheung E, Olin- Sandoval V, Gruning N, Kruger A, Tauqeer Alam M, Keller MA, Breitenbach M, Brindle KM, Rabinowitz JD, and Ralser M The return of metabolism: biochemistry and physiology of the pentose phosphate pathway. Biol Rev Camb Philos Soc (2014).
(82) Jiang P, Du W, and Wu M Regulation of the pentose phosphate pathway in cancer. Protein Cell 5: 592-602 (2014).
(83) Riethdorf S, Reimers N, Assmann V, Kornfeld JW, Terracciano L, Sauter G, and Pantel K High incidence of EMMPRIN expression in human tumors. Int J Cancer 119: 1800- 1810 (2006).
(84) Iacono KT, Brown AL, Greene MI, and Saouaf SJ CD147 immunoglobulin superfamily receptor function and role in pathology. Exp Mol Pathol 83: 283-295 (2007).
(85) Yurchenko V, Constant S, and Bukrinsky M Dealing with the family: CD147 interactions with cyclophilins. Immunology 117: 301-309 (2006).
(86) Muramatsu T, and Miyauchi T Basigin (CD147): a multifunctional transmembrane protein involved in reproduction, neural function, inflammation and tumor invasion. Histol Histopathol 18: 981-987 (2003).
(87) Kaname T, Miyauchi T, Kuwano A, Matsuda Y, Muramatsu T, and Kajii T Mapping basigin (BSG), a member of the immunoglobulin superfamily, to 19p13.3. Cytogenet Cell Genet 64: 195-197 (1993).
154
(88) Beesley PW, Herrera-Molina R, Smalla KH, and Seidenbecher C The Neuroplastin adhesion molecules: key regulators of neuronal plasticity and synaptic function. J Neurochem 131: 268-283 (2014).
(89) Weidle UH, Scheuer W, Eggle D, Klostermann S, and Stockinger H Cancer-related issues of CD147. Cancer Genomics Proteomics 7: 157-169 (2010).
(90) Tang W, Chang SB, and Hemler ME Links between CD147 function, glycosylation, and caveolin-1. Mol Biol Cell 15: 4043-4050 (2004).
(91) Tang W, and Hemler ME Caveolin-1 regulates matrix metalloproteinases-1 induction and CD147/EMMPRIN cell surface clustering. The Journal of biological chemistry 279: 11112-11118 (2004).
(92) Redzic JS, Armstrong GS, Isern NG, Jones DN, Kieft JS, and Eisenmesser EZ The retinal specific CD147 Ig0 domain: from molecular structure to biological activity. Journal of molecular biology 411: 68-82 (2011).
(93) Tang W, Chang SB, and Hemler ME Links between CD147 function, glycosylation, and caveolin-1. Molecular Biology of the Cell 15: 4043-4050 (2004).
(94) Huang W, Luo WJ, Zhu P, Tang J, Yu XL, Cui HY, Wang B, Zhang Y, Jiang JL, and Chen ZN Modulation of CD147-induced matrix metalloproteinase activity: role of CD147 N-glycosylation. The Biochemical journal 449: 437-448 (2013).
(95) Muramatsu T Basigin (CD147), a multifunctional transmembrane glycoprotein with various binding partners. J Biochem (2015).
(96) Yoshida S, Shibata M, Yamamoto S, Hagihara M, Asai N, Takahashi M, Mizutani S, Muramatsu T, and Kadomatsu K Homo-oligomer formation by basigin, an immunoglobulin superfamily member, via its N-terminal immunoglobulin domain. Eur J Biochem 267: 4372-4380 (2000).
(97) Redzic JS, Kendrick AA, Bahmed K, Dahl KD, Pearson CG, Robinson WA, Robinson SE, Graner MW, and Eisenmesser EZ Extracellular vesicles secreted from cancer cell lines stimulate secretion of MMP-9, IL-6, TGF-beta1 and EMMPRIN. PloS one 8: e71225 (2013).
(98) Tang Y, Kesavan P, Nakada MT, and Yan L Tumor-stroma interaction: Positive feedback regulation of extracellular matrix metalloproteinase inducer (EMMPRIN) expression and matrix metalloproteinase-dependent generation of soluble EMMPRIN. Molecular Cancer Research 2: 73-80 (2004).
(99) Schlegel J, Redzic JS, Porter C, Yurchenko V, Bukrinsky M, Armstrong GS, Zhang FL, Isern NG, DeGregori J, Hodges R, and Eisenmesser EZ Solution characterization of the extracellular region of CD147 and its interaction with its enzyme ligand cyclophilin-A. . Journal of Molecular Biology 391: 518-535 (2009).
155
(100) Egawa N, Koshikawa N, Tomari T, Nabeshima K, Isobe T, and Seiki M Membrane type 1 matrix metalloproteinase (MT1-MMP/MMP-14) cleaves and releases a 22-kDa extracellular matrix metalloproteinase inducer (EMMPRIN) fragment from tumor cells. Journal of Biological Chemistry 281: 37576-37585 (2006).
(101) Guo HM, Li RS, Zucker S, and Toole BP EMMPRIN (CD147), an inducer of matrix metalloproteinase synthesis, also binds interstitial collagenase to the tumor cell surface. Cancer Research 60: 888-891 (2000).
(102) Haug C, Lenz C, Diaz F, and Bachem MG Oxidized low-density lipoproteins stimulate extracellular matrix metalloproteinase inducer (EMMPRIN) release by coronary smooth muscle cells. Arteriosclerosis Thrombosis and Vascular Biology 24: 1823- 1829 (2004).
(103) Belton RJ, Jr., Chen L, Mesquita FS, and Nowak RA Basigin-2 is a cell surface receptor for soluble basigin ligand. The Journal of biological chemistry 283: 17805- 17814 (2008).
(104) Bi J, Li Y, Sun F, Saalbach A, Klein C, Miller DJ, Hess R, and Nowak RA Basigin null mutant male mice are sterile and exhibit impaired interactions between germ cells and Sertoli cells. Developmental biology 380: 145-156 (2013).
(105) Kuno N, Kadomatsu K, Fan QW, Hagihara M, Senda T, Mizutani S, and Muramatsu T Female sterility in mice lacking the basigin gene, which encodes a transmembrane glycoprotein belonging to the immunoglobulin superfamily. FEBS letters 425: 191- 194 (1998).
(106) Kasinrerk W, Tokrasinwit N, and Phunpae P CD147 monoclonal antibodies induce homotypic cell aggregation of monocytic cell line U937 via LFA-1/ICAM-1 pathway. Immunology 96: 184-192 (1999).
(107) Sidhu SS, Nawroth R, Retz M, Lemjabbar-Alaoui H, Dasari V, and Basbaum C EMMPRIN regulates the canonical Wnt/beta-catenin signaling pathway, a potential role in accelerating lung tumorigenesis. Oncogene 29: 4145-4156 (2010).
(108) Qin Z, Dai L, Bratoeva M, Slomiany MG, Toole BP, and Parsons C Cooperative roles for emmprin and LYVE-1 in the regulation of chemoresistance for primary effusion lymphoma. Leukemia 25: 1598-1609 (2011).
(109) Zhou S, Liao L, Chen C, Zeng W, Liu S, Su J, Zhao S, Chen M, Kuang Y, Chen X, and Li J CD147 mediates chemoresistance in breast cancer via ABCG2 by affecting its cellular localization and dimerization. Cancer Lett 337: 285-292 (2013).
(110) Pan Y, He B, Song G, Bao Q, Tang Z, Tian F, and Wang S CD147 silencing via RNA interference reduces tumor cell invasion, metastasis and increases chemosensitivity in pancreatic cancer cells. Oncol Rep 27: 2003-2009 (2012).
(111) Schneiderhan W, Scheler M, Holzmann KH, Marx M, Gschwend JE, Bucholz M, Gress TM, Seufferlein T, Adler G, and Oswald F CD147 silencing inhibits lactate transport and reduces malignant potential of pancreatic cancer cells in in vivo and in vitro models. Gut 58: 1391-1398 (2009).
156
(112) Grass GD, Tolliver LB, Bratoeva M, and Toole BP CD147, CD44, and the epidermal growth factor receptor (EGFR) signaling pathway cooperate to regulate breast epithelial cell invasiveness. The Journal of biological chemistry 288: 26089-26104 (2013).
(113) Slomiany MG, Grass GD, Robertson AD, Yang XY, Maria BL, Beeson C, and Toole BP Hyaluronan, CD44, and emmprin regulate lactate efflux and membrane localization of monocarboxylate transporters in human breast carcinoma cells. Cancer Res 69: 1293-1301 (2009).
(114) Toole BP, and Slomiany MG Hyaluronan, CD44 and Emmprin: partners in cancer cell chemoresistance. Drug Resist Updat 11: 110-121 (2008).
(115) Zhou S, Zhou H, Walian PJ, and Jap BK CD147 is a regulatory subunit of the gamma- secretase complex in Alzheimer's disease amyloid beta-peptide production. Proceedings of the National Academy of Sciences of the United States of America 102: 7499-7504 (2005).
(116) Moreno V, Gonzalo P, Gomez-Escudero J, Pollan A, Acin-Perez R, Breckenridge M, Yanez-Mo M, Barreiro O, Orsenigo F, Kadomatsu K, Chen CS, Enriquez JA, Dejana E, Sanchez-Madrid F, and Arroyo AG An EMMPRIN-gamma-catenin-Nm23 complex drives ATP production and actomyosin contractility at endothelial junctions. J Cell Sci 127: 3768-3781 (2014).
(117) Till A, Rosenstiel P, Brautigam K, Sina C, Jacobs G, Oberg HH, Seegert D, Chakraborty T, and Schreiber S A role for membrane-bound CD147 in NOD2- mediated recognition of bacterial cytoinvasion. J Cell Sci 121: 487-495 (2008).
(118) Yurchenko V, Zybarth G, O'Connor M, Dai WW, Franchin G, Hao T, Guo H, Hung HC, Toole B, Gallay P, Sherry B, and Bukrinsky M Active site residues of cyclophilin A are crucial for its signaling activity via CD147. The Journal of biological chemistry 277: 22959-22965 (2002).
(119) Song F, Zhang X, Ren XB, Zhu P, Xu J, Wang L, Li YF, Zhong N, Ru Q, Zhang DW, Jiang JL, Xia B, and Chen ZN Cyclophilin A (CyPA) induces chemotaxis independent of its peptidylprolyl cis-trans isomerase activity: direct binding between CyPA and the ectodomain of CD147. The Journal of biological chemistry 286: 8197-8203 (2011).
(120) Wang P, and Heitman J The cyclophilins. Genome Biol 6: 226 (2005).
(121) Najyb O, Brissette L, and Rassart E Apolipoprotein D Internalization Is a Basigin- dependent Mechanism. The Journal of biological chemistry 290: 16077-16087 (2015).
(122) Hibino T, Sakaguchi M, Miyamoto S, Yamamoto M, Motoyama A, Hosoi J, Shimokata T, Ito T, Tsuboi R, and Huh NH S100A9 is a novel ligand of EMMPRIN that promotes melanoma metastasis. Cancer Res 73: 172-183 (2013).
157
(123) Seizer P, Borst O, Langer HF, Bultmann A, Munch G, Herouy Y, Stellos K, Kramer B, Bigalke B, Buchele B, Bachem MG, Vestweber D, Simmet T, Gawaz M, and May AE EMMPRIN (CD147) is a novel receptor for platelet GPVI and mediates platelet rolling via GPVI-EMMPRIN interaction. Thromb Haemost 101: 682-686 (2009).
(124) Xu D, and Hemler ME Metabolic activation-related CD147-CD98 complex. Mol Cell Proteomics 4: 1061-1071 (2005).
(125) Kirk P, Wilson MC, Heddle C, Brown MH, Barclay AN, and Halestrap AP CD147 is tightly associated with lactate transporters MCT1 and MCT4 and facilitates their cell surface expression. EMBO J 19: 3896-3904 (2000).
(126) Halestrap AP, and Meredith D The SLC16 gene family-from monocarboxylate transporters (MCTs) to aromatic amino acid transporters and beyond. Pflugers Arch 447: 619-628 (2004).
(127) Lin L, Yee SW, Kim RB, and Giacomini KM SLC transporters as therapeutic targets: emerging opportunities. Nature reviews. Drug discovery 14: 543-560 (2015).
(128) Le Floch R, Chiche J, Marchiq I, Naiken T, Ilc K, Murray CM, Critchlow SE, Roux D, Simon MP, and Pouyssegur J CD147 subunit of lactate/H+ symporters MCT1 and hypoxia-inducible MCT4 is critical for energetics and growth of glycolytic tumors. Proceedings of the National Academy of Sciences of the United States of America 108: 16663-16668 (2011).
(129) Baek G, Tse YF, Hu Z, Cox D, Buboltz N, McCue P, Yeo CJ, White MA, DeBerardinis RJ, Knudsen ES, and Witkiewicz AK MCT4 defines a glycolytic subtype of pancreatic cancer with poor prognosis and unique metabolic dependencies. Cell Rep 9: 2233- 2249 (2014).
(130) Aricescu AR, Lu W, and Jones EY A time- and cost-efficient system for high-level protein production in mammalian cells. Acta crystallographica. Section D, Biological crystallography 62: 1243-1250 (2006).
(131) Bos AB, Luan P, Duque JN, Reilly D, Harms PD, and Wong AW Optimization and automation of an end-to-end high throughput microscale transient protein production process. Biotechnology and bioengineering 112: 1832-1842 (2015).
(132) Chaudhary S, Pak JE, Gruswitz F, Sharma V, and Stroud RM Overexpressing human membrane proteins in stably transfected and clonal human embryonic kidney 293S cells. Nature protocols 7: 453-466 (2012).
(133) Gu J, Gao X, Pan X, Peng X, Li Y, and Li M High-level expression and one-step purification of a soluble recombinant human interleukin-37b in Escherichia coli. Protein Expr Purif 108: 18-22 (2015).
(134) Wurm F, and Bernard A Large-scale transient expression in mammalian cells for recombinant protein production. Current opinion in biotechnology 10: 156-159 (1999).
158
(135) Stowell SR, Ju T, and Cummings RD Protein glycosylation in cancer. Annual review of pathology 10: 473-510 (2015).
(136) Deer EL, Gonzalez-Hernandez J, Coursen JD, Shea JE, Ngatia J, Scaife CL, Firpo MA, and Mulvihill SJ Phenotype and genotype of pancreatic cancer cell lines. Pancreas 39: 425-435 (2010).
(137) Bystrykh L, Weersing E, Dontje B, Sutton S, Pletcher MT, Wiltshire T, Su AI, Vellenga E, Wang J, Manly KF, Lu L, Chesler EJ, Alberts R, Jansen RC, Williams RW, Cooke MP, and de Haan G Uncovering regulatory pathways that affect hematopoietic stem cell function using 'genetical genomics'. Nat Genet 37: 225-232 (2005).
(138) Grzesiak JJ, Ho JC, Moossa AR, and Bouvet M The integrin-extracellular matrix axis in pancreatic cancer. Pancreas 35: 293-301 (2007).
(139) He W, Sun Z, and Liu Z Silencing of TGM2 reverses epithelial to mesenchymal transition and modulates the chemosensitivity of breast cancer to docetaxel. Exp Ther Med 10: 1413-1418 (2015).
(140) Hosper NA, van den Berg PP, de Rond S, Popa ER, Wilmer MJ, Masereeuw R, and Bank RA Epithelial-to-mesenchymal transition in fibrosis: collagen type I expression is highly upregulated after EMT, but does not contribute to collagen deposition. Exp Cell Res 319: 3000-3009 (2013).
(141) Marcato P, Dean CA, Pan D, Araslanova R, Gillis M, Joshi M, Helyer L, Pan L, Leidal A, Gujar S, Giacomantonio CA, and Lee PW Aldehyde dehydrogenase activity of breast cancer stem cells is primarily due to isoform ALDH1A3 and its expression is predictive of metastasis. Stem Cells 29: 32-45 (2011).
(142) Jia J, Bosley AD, Thompson A, Hoskins JW, Cheuk A, Collins I, Parikh H, Xiao Z, Ylaya K, Dzyadyk M, Cozen W, Hernandez BY, Lynch CF, Loncarek J, Altekruse SF, Zhang L, Westlake CJ, Factor VM, Thorgeirsson S, Bamlet WR, Hewitt SM, Petersen GM, Andresson T, and Amundadottir LT CLPTM1L promotes growth and enhances aneuploidy in pancreatic cancer cells. Cancer Res 74: 2785-2795 (2014).
(143) Ni XG, Zhou L, Wang GQ, Liu SM, Bai XF, Liu F, Peppelenbosch MP, and Zhao P The ubiquitin-proteasome pathway mediates gelsolin protein downregulation in pancreatic cancer. Mol Med 14: 582-589 (2008).
(144) Hayes MJ, Rescher U, Gerke V, and Moss SE Annexin-actin interactions. Traffic 5: 571-576 (2004).
(145) Maschler S, Gebeshuber CA, Wiedemann EM, Alacakaptan M, Schreiber M, Custic I, and Beug H Annexin A1 attenuates EMT and metastatic potential in breast cancer. EMBO Mol Med 2: 401-414 (2010).
(146) Ullah MS, Davies AJ, and Halestrap AP The plasma membrane lactate transporter MCT4, but not MCT1, is up-regulated by hypoxia through a HIF-1alpha-dependent mechanism. The Journal of biological chemistry 281: 9030-9037 (2006).
159
(147) Ovens MJ, Davies AJ, Wilson MC, Murray CM, and Halestrap AP AR-C155858 is a potent inhibitor of monocarboxylate transporters MCT1 and MCT2 that binds to an intracellular site involving transmembrane helices 7-10. The Biochemical journal 425: 523-530 (2010).
(148) Zhao C, Wilson MC, Schuit F, Halestrap AP, and Rutter GA Expression and distribution of lactate/monocarboxylate transporter isoforms in pancreatic islets and the exocrine pancreas. Diabetes 50: 361-366 (2001).
(149) Trnka MJ, Baker PR, Robinson PJ, Burlingame AL, and Chalkley RJ Matching cross- linked peptide spectra: only as good as the worse identification. Mol Cell Proteomics 13: 420-434 (2014).
(150) James AD, Chan A, Erice O, Siriwardena AK, and Bruce JI Glycolytic ATP fuels the plasma membrane calcium pump critical for pancreatic cancer cell survival. The Journal of biological chemistry 288: 36007-36019 (2013).
(151) James AD, Patel W, Butt Z, Adiamah M, Dakhel R, Latif A, Uggenti C, Swanton E, Imamura H, Siriwardena AK, and Bruce JI The Plasma Membrane Calcium Pump in Pancreatic Cancer Cells Exhibiting the Warburg Effect Relies on Glycolytic ATP. The Journal of biological chemistry 290: 24760-24771 (2015).
(152) Supper V, Schiller HB, Paster W, Forster F, Boulegue C, Mitulovic G, Leksa V, Ohradanova-Repic A, Machacek C, Schatzlmaier P, Zlabinger GJ, and Stockinger H Association of CD147 and Calcium Exporter PMCA4 Uncouples IL-2 Expression from Early TCR Signaling. J Immunol 196: 1387-1399 (2016).
(153) Cantor JM, and Ginsberg MH CD98 at the crossroads of adaptive immunity and cancer. J Cell Sci 125: 1373-1382 (2012).
(154) Wilson MC, Meredith D, and Halestrap AP Fluorescence resonance energy transfer studies on the interaction between the lactate transporter MCT1 and CD147 provide information on the topology and stoichiometry of the complex in situ. The Journal of biological chemistry 277: 3666-3672 (2002).
(155) Saibil H Chaperone machines for protein folding, unfolding and disaggregation. Nat Rev Mol Cell Biol 14: 630-642 (2013).
(156) Moreno-Layseca P, and Streuli CH Signalling pathways linking integrins with cell cycle progression. Matrix Biol 34: 144-153 (2014).
(157) Sinha A, Cherba D, Bartlam H, Lenkiewicz E, Evers L, Barrett MT, and Haab BB Mesenchymal-like pancreatic cancer cells harbor specific genomic alterations more frequently than their epithelial-like counterparts. Molecular oncology 8: 1253-1265 (2014).
(158) Li L, Tang W, Wu X, Karnak D, Meng X, Thompson R, Hao X, Li Y, Qiao XT, Lin J, Fuchs J, Simeone DM, Chen ZN, Lawrence TS, and Xu L HAb18G/CD147 promotes pSTAT3-mediated pancreatic cancer development via CD44s. Clinical cancer research : an official journal of the American Association for Cancer Research 19: 6703-6715 (2013).
160
(159) Tempia-Caliera AA, Horvath LZ, Zimmermann A, Tihanyi TT, Korc M, Friess H, and Buchler MW Adhesion molecules in human pancreatic cancer. J Surg Oncol 79: 93- 100 (2002).
(160) Prevarskaya N, Skryma R, and Shuba Y Calcium in tumour metastasis: new roles for known actors. Nat Rev Cancer 11: 609-618 (2011).
(161) Park CJ, and Seo YS Heat Shock Proteins: A Review of the Molecular Chaperones for Plant Immunity. Plant Pathol J 31: 323-333 (2015).
(162) Whitley D, Goldberg SP, and Jordan WD Heat shock proteins: a review of the molecular chaperones. J Vasc Surg 29: 748-751 (1999).
(163) Whitesell L, and Lindquist SL HSP90 and the chaperoning of cancer. Nat Rev Cancer 5: 761-772 (2005).
(164) Grammatikakis N, Vultur A, Ramana CV, Siganou A, Schweinfest CW, Watson DK, and Raptis L The role of Hsp90N, a new member of the Hsp90 family, in signal transduction and neoplastic transformation. The Journal of biological chemistry 277: 8312-8320 (2002).
(165) Bergeron JJ, Brenner MB, Thomas DY, and Williams DB Calnexin: a membrane- bound chaperone of the endoplasmic reticulum. Trends Biochem Sci 19: 124-128 (1994).
(166) Kremer CL, Schmelz M, and Cress AE Integrin-dependent amplification of the G2 arrest induced by ionizing radiation. Prostate 66: 88-96 (2006).
(167) Howard BR, Vajdos FF, Li S, Sundquist WI, and Hill CP Structural insights into the catalytic mechanism of cyclophilin A. Nat Struct Biol 10: 475-481 (2003).
(168) Nigro P, Pompilio G, and Capogrossi MC Cyclophilin A: a key player for human disease. Cell Death Dis 4: e888 (2013).
(169) Schiene C, and Fischer G Enzymes that catalyse the restructuring of proteins. Curr Opin Struct Biol 10: 40-45 (2000).
(170) Shaw ML, Stone KL, Colangelo CM, Gulcicek EE, and Palese P Cellular proteins in influenza virus particles. PLoS Pathog 4: e1000085 (2008).
(171) Chatterji U, Bobardt M, Selvarajah S, Yang F, Tang H, Sakamoto N, Vuagniaux G, Parkinson T, and Gallay P The isomerase active site of cyclophilin A is critical for hepatitis C virus replication. J Biol Chem 284: 16998-17005 (2009).
(172) Patient R, Hourioux C, Sizaret PY, Trassard S, Sureau C, and Roingeard P Hepatitis B virus subviral envelope particle morphogenesis and intracellular trafficking. J Virol 81: 3842-3851 (2007).
(173) Towers GJ, Hatziioannou T, Cowan S, Goff SP, Luban J, and Bieniasz PD Cyclophilin A modulates the sensitivity of HIV-1 to host restriction factors. Nat Med 9: 1138-1143 (2003).
161
(174) Sanchez-Tillo E, Wojciechowska M, Comalada M, Farrera C, Lloberas J, and Celada A Cyclophilin A is required for M-CSF-dependent macrophage proliferation. Eur J Immunol 36: 2515-2524 (2006).
(175) Syed F, Rycyzyn MA, Westgate L, and Clevenger CV A novel and functional interaction between cyclophilin A and prolactin receptor. Endocrine 20: 83-90 (2003).
(176) Andreotti AH Native state proline isomerization: an intrinsic molecular switch. Biochemistry 42: 9515-9524 (2003).
(177) Lee J Role of cyclophilin a during oncogenesis. Archives of pharmacal research 33: 181-187 (2010).
(178) Colgan J, Asmal M, Yu B, and Luban J Cyclophilin A-deficient mice are resistant to immunosuppression by cyclosporine. J Immunol 174: 6030-6038 (2005).
(179) Arora K, Gwinn WM, Bower MA, Watson A, Okwumabua I, MacDonald HR, Bukrinsky MI, and Constant SL Extracellular cyclophilins contribute to the regulation of inflammatory responses. J Immunol 175: 517-522 (2005).
(180) Suzuki J, Jin ZG, Meoli DF, Matoba T, and Berk BC Cyclophilin A is secreted by a vesicular pathway in vascular smooth muscle cells. Circ Res 98: 811-817 (2006).
(181) Xu Q, Leiva MC, Fischkoff SA, Handschumacher RE, and Lyttle CR Leukocyte chemotactic activity of cyclophilin. The Journal of biological chemistry 267: 11968- 11971 (1992).
(182) Sherry B, Yarlett N, Strupp A, and Cerami A Identification of Cyclophilin as a Proinflammatory Secretory Product of Lipopolysaccharide-Activated Macrophages. Proceedings of the National Academy of Sciences of the United States of America 89: 3511-3515 (1992).
(183) Bahmed K, Henry C, Holliday M, Redzic J, Ciobanu M, Zhang F, Weekes C, Sclafani R, Degregori J, and Eisenmesser E Extracellular cyclophilin-A stimulates ERK1/2 phosphorylation in a cell-dependent manner but broadly stimulates nuclear factor kappa B. Cancer Cell Int 12: 19 (2012).
(184) Saphire AC, Bobardt MD, and Gallay PA Host cyclophilin A mediates HIV-1 attachment to target cells via heparans. EMBO J 18: 6771-6785 (1999).
(185) Li M, Zhai Q, Bharadwaj U, Wang H, Li F, Fisher WE, Chen C, and Yao Q Cyclophilin A is overexpressed in human pancreatic cancer cells and stimulates cell proliferation through CD147. Cancer 106: 2284-2294 (2006).
(186) Dmochewitz L, Lillich M, Kaiser E, Jennings LD, Lang AE, Buchner J, Fischer G, Aktories K, Collier RJ, and Barth H Role of CypA and Hsp90 in membrane translocation mediated by anthrax protective antigen. Cell Microbiol 13: 359-373 (2011).
(187) Polo S, Pece S, and Di Fiore PP Endocytosis and cancer. Curr Opin Cell Biol 16: 156- 161 (2004).
162
(188) Burke P, Schooler K, and Wiley HS Regulation of epidermal growth factor receptor signaling by endocytosis and intracellular trafficking. Mol Biol Cell 12: 1897-1910 (2001).
(189) Sorkin A, and von Zastrow M Endocytosis and signalling: intertwining molecular networks. Nat Rev Mol Cell Biol 10: 609-622 (2009).
(190) Doherty GJ, and McMahon HT Mechanisms of endocytosis. Annual review of biochemistry 78: 857-902 (2009).
(191) Lim JP, and Gleeson PA Macropinocytosis: an endocytic pathway for internalising large gulps. Immunol Cell Biol 89: 836-843 (2011).
(192) Flannagan RS, Jaumouille V, and Grinstein S The cell biology of phagocytosis. Annual review of pathology 7: 61-98 (2012).
(193) Commisso C, Davidson SM, Soydaner-Azeloglu RG, Parker SJ, Kamphorst JJ, Hackett S, Grabocka E, Nofal M, Drebin JA, Thompson CB, Rabinowitz JD, Metallo CM, Vander Heiden MG, and Bar-Sagi D Macropinocytosis of protein is an amino acid supply route in Ras-transformed cells. Nature 497: 633-637 (2013).
(194) Bruns CJ, Harbison MT, Kuniyasu H, Eue I, and Fidler IJ In vivo selection and characterization of metastatic variants from human pancreatic adenocarcinoma by using orthotopic implantation in nude mice. Neoplasia 1: 50-62 (1999).
(195) Akiyama T, Ishida J, Nakagawa S, Ogawara H, Watanabe S, Itoh N, Shibuya M, and Fukami Y Genistein, a specific inhibitor of tyrosine-specific protein kinases. The Journal of biological chemistry 262: 5592-5595 (1987).
(196) Yang JM, O'Neill P, Jin W, Foty R, Medina DJ, Xu Z, Lomas M, Arndt GM, Tang Y, Nakada M, Yan L, and Hait WN Extracellular matrix metalloproteinase inducer (CD147) confers resistance of breast cancer cells to Anoikis through inhibition of Bim. The Journal of biological chemistry 281: 9719-9727 (2006).
(197) Saleh T, Jankowski W, Sriram G, Rossi P, Shah S, Lee KB, Cruz LA, Rodriguez AJ, Birge RB, and Kalodimos CG Cyclophilin A promotes cell migration via the Abl-Crk signaling pathway. Nat Chem Biol 12: 117-123 (2016).
(198) Holliday MJ, Zhang F, Isern NG, Armstrong GS, and Eisenmesser EZ 1H, 13C, and 15N backbone and side chain resonance assignments of thermophilic Geobacillus kaustophilus cyclophilin-A. Biomol NMR Assign 8: 23-27 (2014).
(199) Eisenmesser EZ, Bosco DA, Akke M, and Kern D Enzyme dynamics during catalysis. Science 295: 1520-1523 (2002).
(200) Meng L, Forouhar F, Thieker D, Gao Z, Ramiah A, Moniz H, Xiang Y, Seetharaman J, Milaninia S, Su M, Bridger R, Veillon L, Azadi P, Kornhaber G, Wells L, Montelione GT, Woods RJ, Tong L, and Moremen KW Enzymatic basis for N-glycan sialylation: structure of rat alpha2,6-sialyltransferase (ST6GAL1) reveals conserved and unique features for glycan sialylation. The Journal of biological chemistry 288: 34680-34698 (2013).
163
(201) Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J, and Bax A NMRPipe: a multidimensional spectral processing system based on UNIX pipes. Journal of biomolecular NMR 6: 277-293 (1995).
(202) Vranken WF, Boucher W, Stevens TJ, Fogh RH, Pajon A, Llinas M, Ulrich EL, Markley JL, Ionides J, and Laue ED The CCPN data model for NMR spectroscopy: development of a software pipeline. Proteins 59: 687-696 (2005).
(203) Loria JP, Rance M, and Palmer AG, 3rd A TROSY CPMG sequence for characterizing chemical exchange in large proteins. Journal of biomolecular NMR 15: 151-155 (1999).
(204) Cavalli A, Camilloni C, and Vendruscolo M Molecular dynamics simulations with replica-averaged structural restraints generate structural ensembles according to the maximum entropy principle. The Journal of chemical physics 138: 094112 (2013).
(205) Pronk S, Pall S, Schulz R, Larsson P, Bjelkmar P, Apostolov R, Shirts MR, Smith JC, Kasson PM, van der Spoel D, Hess B, and Lindahl E GROMACS 4.5: a high- throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 29: 845-854 (2013).
(206) Bonomi M, Branduardi D, Bussi G, Camilloni C, Provasi D, Raiteri P, Donadio D, Marinelli F, Pietrucci F, Broglia RA, and Parrinello M PLUMED: A portable plugin for free-energy calculations with molecular dynamics. Comput Phys Commun 180: 1961-1972 (2009).
(207) Cavalli A, Vendruscolo M, and Paci E Comparison of sequence-based and structure- based energy functions for the reversible folding of a peptide. Biophysical journal 88: 3158-3166 (2005).
(208) Piana S, Lindorff-Larsen K, and Shaw DE How robust are protein folding simulations with respect to force field parameterization? Biophysical journal 100: L47-49 (2011).
(209) Hess B P-LINCS: A parallel linear constraint solver for molecular simulation. J Chem Theory Comput 4: 116-122 (2008).
(210) Essmann U, Perera L, Berkowitz ML, Darden T, Lee H, and Pedersen LG A Smooth Particle Mesh Ewald Method. J Chem Phys 103: 8577-8593 (1995).
(211) Bussi G, Donadio D, and Parrinello M Canonical sampling through velocity rescaling. J Chem Phys 126 (2007).
(212) Clore GM, Appella E, Yamada M, Matsushima K, and Gronenborn AM Three- dimensional structure of interleukin 8 in solution. Biochemistry 29: 1689-1696 (1990).
(213) DeLano WL, and Lam JW PyMOL: A communications tool for computational models. Abstr Pap Am Chem S 230: U1371-U1372 (2005).
164
(214) Kendrick AA, Choudhury M, Rahman SM, McCurdy CE, Friederich M, Van Hove JL, Watson PA, Birdsey N, Bao J, Gius D, Sack MN, Jing E, Kahn CR, Friedman JE, and Jonscher KR Fatty liver is associated with reduced SIRT3 activity and mitochondrial protein hyperacetylation. The Biochemical journal 433: 505-514 (2011).
(215) Rejman J, Bragonzi A, and Conese M Role of clathrin- and caveolae-mediated endocytosis in gene transfer mediated by lipo- and polyplexes. Mol Ther 12: 468-474 (2005).
(216) Nemkov T, D'Alessandro A, and Hansen KC Three-minute method for amino acid analysis by UHPLC and high-resolution quadrupole orbitrap mass spectrometry. Amino Acids 47: 2345-2357 (2015).
165