Background
Cancer is a prevalent cause of death, and is estimated to have been responsible for 8.2 million deaths globally in 2012. Accounting for approximately 521 000 deaths every year, breast cancer is one of the most common causes of cancer-related death [
1].
Changes in cell metabolism are central to cancer development [
2]. Mitochondria play a central role in regulating parameters of the metabolism, such as energy production, production of biosynthetic precursors, redox status, reactive oxygen species generation, cytosolic calcium levels, and the initiation of apoptosis. In both normal cells and cancer cells, changes in these parameters may prompt a shift in the cell state, e.g., from quiescent and differentiated to actively proliferating. Increases and decreases in mitochondrial activity can be mediated by mutations in genes encoding mitochondrial proteins. In the context of breast cancer, several studies on mitochondrial alterations are available and have been reviewed in [
3]. Only a small fraction of mitochondrial proteins are encoded by the mitochondrial DNA. Nuclear DNA encodes all other mitochondrial proteins, including mitochondrial ribosomal proteins (Mrps). This emphasizes the integrated nature of the cross-talk between the mitochondria and the nucleus.
Early proteomic analysis of the mammalian mitochondrial ribosome reported three variants of Mrps18 [
4,
5]. Although these early studies revealed that members of the s18 protein family are localized on the surface of the small subunit (S28) of the mitochondrial ribosome, the function of these proteins is largely unknown. Recent data reporting the complete structure of the large mitochondrial ribosomal subunit has revealed that Mrps18a is not located on the small subunit as previously indicated, but on the large ribosomal subunit [
6], which demonstrates the lack of knowledge concerning the importance and function of Mrps18a. The three variants, designated Mprs18a, Mrps18b, and Mrps18c, are the products of three separate genes. The genes are only 25–30% identical, which is comparable to their homology with their bacterial counterparts [
4]. It has been suggested that there may be different sub-populations of cells within tissues where the expression levels of the different Mrps18 variants vary [
7]. In 2009, it was shown that overexpression of human Mrps18b in primary rat embryonic fibroblasts leads to immortalization. A cell line derived from these cells, called 18IM, adopted several cancer-like characteristics, such as loss of contact inhibition and anchorage-independent growth. Furthermore, 18IM cells acquired expression of different stem cell markers [
8]. Microarray and Q-PCR analysis of 18IM and three other equivalent cell lines showed up-regulation of a range of genes involved in mitochondrial pathways such as energy production, which are characteristic of rapidly proliferating cells [
9]. Recently, a study of overexpression of Mrps18b in human breast and renal cancer cell lines has shown that Mrps18b results in the appearance of multinucleated cells [
10]. These observations indicate that Mrps18 family proteins, or at least Mrps18b, may be involved in neoplastic transformation.
It is widely believed that intertumor heterogeneity reflects the “cell of origin” as well as the mutational profile of a given cancer cell [
11,
12]. Much effort has been put into unravelling the normal cellular hierarchy of the human breast, as it is an important step toward understanding the “cells of origin” and molecular events that drive breast cancer. Several populations of cells within the human breast have been proposed as candidates for the “cells of origin” of cancer, including stem cells and progenitors, or transit-amplifying cells. One such candidate population was identified in a study performed by Villadsen et al., and supported by Honeth et al. [
13], where a putative stem cell zone was identified in the ducts of the normal human breast containing multipotent cells positive for both the luminal epithelial lineage marker, cytokeratin 19 (K19), and the myoepithelial lineage marker, cytokeratin 14 (K14) [
14].
In this study, we aimed to identify proteins which are differentially regulated in breast cancer and may therefore have a potential use in the diagnosis and/or treatment of breast cancer. Phage display-mediated selections of human domain antibodies were performed against ductal regions containing K19
+/K14
+ cells in normal breast tissue cryostat sections using the shadow stick selection technique [
15,
16] adopted from [
17‐
19]. Here, a recombinant antibody library, displayed in genetic fusion with the filamentous bacteriophage protein 3 [
20], is allowed to bind to antigens presented on the tissue sections. Compared to selection on intact cells obtained from circulation or cell culture, the tissue sections allow all cellular compartments to be targeted, as the sections normally have a thickness of 4–8 μM. Following removal of non-binding phage antibodies by washing, the target cells in the tissue are covered by a shadow stick and the tissue section is illuminated by UV-C light. The UV-C light causes crosslinking in the phage genome, resulting in phage particles incapable of replication after infection. Phage particles covered by the shadow stick are protected from the UV-C light and remain able to replicate after infection. In this way, phage display selection becomes a discovery tool in the search for proteins which are differentially regulated between distinct cell populations, which may represent novel biomarkers.
A potentially interesting antigen, Mrps18a, emerged as the cognate antigen of one of the selected phage antibodies. Although the functions of this particular protein are currently unexplored both in the context of the normal cell state as well as in cancer cells, we found that the expression of Mrps18a is upregulated in breast cancer cells compared to normal cells. Mrps18a cannot be considered a unique biomarker for K19 and K14 positive cells, as it is found in both normal and cancer cells. However, its increased expression in cancer cells is explained by the increased energy metabolism of cancer cells, and it may pave the way for new diagnostic and therapeutic routes to be explored.
Methods
Tissue sections
Biopsies of healthy human breast tissue and breast cancer tissue were obtained from five patients (P636, P659, P722, P727, and P819) undergoing reduction mammoplasty and three cancer patients (P757, P942, and P949), respectively. The use of human material has been reviewed by the Regional Scientific Ethical Committees (Region Hovedstaden) and approved with reference to H-2-2011-052 and H-2-2010-051. Informed consent was obtained from each patient regarding the collection and use of tissue. Cryostat sections (8 μm) were prepared as described by [
21]. All tissue sections were fixed in methanol for 5 min at −20 °C.
Target area identification
In preparation for selection, tissue sections were stained by immunohistochemistry (IHC) with antibodies against K19 and K14 and detected with Alexa Flour A568 and A488 conjugated isotype specific goat anti-mouse antibodies, respectively. For details on the antibodies, see Tables
1 and
2.
Table 1
Primary antibodies used for IHC
Mouse monoclonal Anti-K14 | IgG3 | 1:50 | Leica Biosystems, NCL-L-LL002 |
Mouse monoclonal Anti-K19 | IgG2a | 1:100 | Abcam, ab7754 |
dAb | - | ~250 μg/mL | - |
dAb-rFc | - | 40–100 μg/mL | - |
Table 2
Secondary antibodies used for IHC
Goat anti-mouse | A488 | IgG3
| 1:500 | Invitrogen, A21151 | CK14 |
Goat anti-mouse | A350 | IgG2a
| 1:250 | Invitrogen, A21130 | CK19 |
Goat anti-mouse | A488 | IgG2a
| 1:500 | Invitrogen, A21131 | CK19 |
Mouse anti-human c-Myc | Cy3 | IgG1
| 1:250 | Sigma-Aldrich, C6594 | dAb |
Goat anti-rabbit | A488 | IgG (H + L) | 1:400 | Invitrogen, A11008 | dAb-rFc |
One tissue section from each patient was mounted with Fluoromount (Sigma-Aldrich), and areas containing rare K19+/14+ cells were identified. Only patient samples containing 1–2 areas with 2–10 luminal positioned K19+/14+ cells residing in ductal regions were used for selections.
Selection
The tissue section was blocked for 1 h in 4% Marvel dried skim milk powder (MPBS). The tissue section was then incubated with the Predator single domain library [
22] in a slide container containing 20 mL 2% MPBS for 1½ h on a turntable under gentle agitation and for 1½ h without agitation. Approximately 10
12 phage particles of the Predator library were applied [
22]. The tissue section was washed twice for 10 min in PBS and twice for 10 min in PBS with 10% glycerol (PBSG) under gentle agitation. The tissue section was dried except the target area, which was kept moist with PBSG. Using brightfield microscopy, a custom made shadow stick (~80 μm) was positioned above the target area. Shadow sticks were prepared as previously described [
23]. The tissue section was exposed to UV-C light (254 nm) for 10 min using a UV-C source (model UVSL-14P from UVP, Upland) positioned approximately 5 cm above the tissue section. Phage particles were eluted with 15 μL trypsin (1 mg/mL) for 10 min. Trypsin was aspirated and transferred to 50 μL ice cold foetal bovine serum. Subsequently, the target area was washed 15 times with 50 μL PBSG. This liquid was transferred to the eluate before storage overnight at −20 °C. The trypsinated phages were propagated in
E. coli and single colonies were cultured in a 96-well plate as described by [
23].
Phage antibody ELISA
Phage antibodies were produced by superinfection with the KM13 helper phage [
24]. This was performed in 96-well format for screening and in 50 mL cultures for ELISA with dilution arrays of monoclonal phage antibodies respectively followed by precipitation with 20% poly ethylene glycol (PEG) 6000, 2.5 M NaCl. Phage titers from 50 mL cultures were estimated by calculation using absorbance measured at 269 and 320 nm as described by [
25].
Costar 96 well culture plates (Corning) with luminal breast cells enriched for K19+/K14+ cells were prepared by FACS and short term culturing. Luminal cells were initially sorted from normal breast tissue (source as describes for tissue sections) with EpCAM, then cultured in FAD2 medium consisting of Ham’s F12 : DMEM mixed in a 1:3 ratio (Tan DW, 2013), and 5% FCS (Liu X, 2012). The cells were cultured on a feeder layer of mouse fibroblasts to minimize the arisal of basal K14+ cells. Sorting of EpCAM+/CD271+ cells from these cultures resulted in a population enriched for K19+/K14+ cells. These were cultured in FAD2 medium in collagen coated flasks. K19+/K14+ and K19+ cells were represented approximately 1:1. The cells were seeded in 96-wells, approximately 10–20 000 cells/well, and cultured 3–5 days before fixation for 10 min with 3.7% formalin, followed by 10 min with 0.1% triton X-100. The plates were dried overnight at 37 °C.
For screening, 50 μl 4% MPBS and 50 μL of the individual PEG precipitated phage antibodies were transfered to a cell coated 96-well plate. For ELISA with dilution arrays of monoclonal phage antibodies, a six pointed 10-fold dilution series was prepared. From this, 100 μL was transferred to the 96-well plate resulting in concentrations ranging from 10
11 phages/well to 10
7 phages/well. As a positive control, the phage scFv antibody 52 was used [
26]. This phage antibody has been reported to have affinity towards at least 14 different cell lines, among these breast cancer cell line MCF7 [
26]. As a negative control, a phage scFv antibody specific to foetal Epsilon-haemoglobin, DAb1, was included [
27]. For both screening and ELISA with dilution arrays, bound phage antibodies were detected with HRP conjugated M13 phage antibody (GE healthcare) 1:3000 and TMB Plus “Ready-to-use” substrate solution (Thermo, Fischer Scientific) according to the manufacturer’s protocol. The chromogenic reaction was stopped after 10–15 min by addition of 50 μL 1 M H
2SO
4. Absorbance (OD
450) was read on a Bio-Rad Model 550 microplate reader (Bio-Rad) with subtraction of background (OD
655).
Expression of single domain antibody (dAb) in E. coli
For expression of soluble dAb, the sequence encoding BC5 was sub-cloned from phagemid into pET22b-MycHis, a modified version of pET22b (Merck, KGaA), using NcoI and NotI (FastDigest, Fermentas) according to the manufacturer’s protocol. This expression plasmid was transformed into BL21 Gold DE3 (Aglient Technologies). For expression, 200 mL exponentially growing cultures were set up in TB growth medium supplemented with 100 μg/mL ampicillin and 4% w/v glucose at 37 °C, 200 rpm. At an OD600 of approximately 0.6–0.8, protein expression was induced by exchange of growth medium to TB containing 100 μg/mL ampicillin and 1 μM IPTG and incubated 18–20 h at 18 °C, 200 rpm. Protein was precipitated using 30% w/v ammonium sulphate. Pelleted protein was resuspended in 20 mL Protein A binding buffer TBS (50 mM Tris, 150 mM NaCl) pH 8.0 added 0.1 mM DNase (Roche). The protein suspension was sterile filtered (0.45 μm) before affinity purification on HiTrap Protein A column (GE Healthcare) according to the manufacturer’s protocol, except eluted with 2.5% acetic acid. Purified antibody fragments were transferred to TBS, pH 7 by dialysis. A purity over 95% was confirmed by SDS-PAGE, and the protein concentration was assessed with a NanoDrop 1000 (Thermo, Fisher Scientific).
Expression of dAb with rabbit Fc-domain (dAb-rFc) in L. tarentolae T7-TR
In order to express soluble dAb-rFc, the sequence encoding BC5 antibody was sub-cloned from phagemid into pMJ_LEXSY-rFc vector, and expressed and purified as described [
28].
Immunohistochemistry
For initial IHC, tissue sections were blocked in Ultravision Block (Thermo scientific) for 1 h. Incubation with primary antibody (Table
1) in a total volume of 100 μL Ultra Vision Block with 10% goat serum (Sigma Aldrich) was performed overnight at 4 °C in a humidity box. Incubation with fluorescent secondary antibodies (Table
2) were performed for 30 min in 100 μL Ultra Vision Block with 10% goat serum. Flouromount (Sigma-Aldrich) was used for mounting. A frozen tissue microarray with 5 μm sections (BioChain
®, Cat. T6235086-1) was blocked and stained as described above. Fluorescence microscopy was performed with a Leica DMI 3000 B microscope with Cell B imaging software.
Antigen identification–protein macroarray
Screening of a UniPEx1 protein macroarray (BioScience ImaGenes) was performed according to the manufacturer’s protocol with purified recombinant antibody, either dAb or dAb-rFc. Briefly, the activated membrane was blocked with 5% BSA for 1 h before incubation overnight with 250 μg dAb or 125 μg dAb-rFc. Detection was performed by incubation for 2 h with secondary antibody; Monoclonal Anti-c-Myc − Cy3™ (Sigma-Aldrich) 1:2000 or Goat anti-Rabbit-IgG-Cy5 (Abcam, ab6564) 1:8000. The membrane was scanned with a Typhoon TRIO variable mode imager (Amersham Biosciences). Antibody binding to the proteins on the array was evaluated using an in-house developed Matlab program. The immobilised proteins were situated on the membrane in defined duplicated patterns arranged in squares consisting of 3x3 spots with a central ink dot. By evaluating the intensities of different spots compared to the background, as well as their positions, the program created a list of possible antigen hits listed from highest to lowest intensity. Only hits detected with both dAb and dAb-rFc formats were considered. Furthermore, hits detected only once despite being spotted several times in duplex were omitted.
Indirect ELISA with Mrps18a
MRPS18A cDNA was purchased (Source BioScience, IRAUp969D0453D) and the sequence verified. The insert was sub-cloned into pETM11 after PCR amplification using primers inserting recognition sequences for the restriction enzymes NotI and NcoI. This allowed for cytosolic expression in BL21 Gold DE3 and subsequent affinity purification under native conditions on Ni-NTA Superflow Beads (QIAGEN) according to manufacturer’s protocol.
A 96-well MaxiSorp plate (Thermo, Fisher Scientific) was coated with purified Mrps18a in concentrations ranging from 1 to 1500 ng in TBS, pH 7.5, per well and blocked with 2% BSA. The plate was incubated for 2.5 h under gentle agitation with 100 μg/mL or 200 μg/mL dAb. Bound antibodies were detected with HRP conjugated Anti-c-Myc (not commercial) 1:400. TMB “Ready-to-use” substrate was applied as described above.
Far western blot analysis
Purified Mrps18a, was run in a standard SDS-PAGE under reducing conditions and then electro blotted onto a PVDF membrane using the traditional sandwich method. Denaturing and renaturing of protein in the membrane was facilitated by subsequent steps of incubation with AC buffers with decreasing concentrations of guanidine–HCl (6 M–0 M) as described [
29]. Subsequently, the membrane was incubated with 10 μg BC5 (dAb-rFc), followed by HRP conjugated swine anti-rabbit (DAKO) 1:4000. Detection was carried out with an ECL detection kit (Amersham Biosciences).
Discussion
This study describes the identification of Mrps18a as an antigen which exhibits increased expression by some human breast cancer cells compared to normal cells. Traditionally, changes in the expression of proteins in cancer cells have been studied using both expression studies looking at the mRNA levels and proteomics using 2D gels and mass spectrometry to identify changes at the proteome level. Here, we have used the selection of recombinant phage antibody libraries as a discovery tool to identify proteins which exhibit differential expression. The putative stem cell niche on which we have selected contained K19
+/K14
+ cells, which are candidate cells of origin of breast cancer [
14]. The principle of the selection method relies on the retrieval of phage-displayed antibody fragments binding to a minute area of interest in a tissue section. Using the shadow stick, only phages binding to the target area were able to replicate in
E. Coli. This method yields a low output of clones from each selection compared to traditional selections, while generating a relative high frequency of phage antibodies binding to unique or upregulated antigens [
23]. Important for this approach is the choice of screening material, i.e., the nature of the cells on which the selected antibodies are tested. When targeting rare cells there is always a compromise between availability of suitable cells and the extent to which they resemble the target cells. In this study, a limiting factor was the availability of primary K19
+/K14
+ cells for screening purposes. Here, short-term-cultured K19
+/K14
+ cells were used for screening. The use of short-term cultivation limited the total number of cells available for the initial ELISA testing, preventing triplicate measurements. The initial ELISA was only used as a means for defining the phage antibodies which were to be considered for further analysis. When performing ELISA with the antibody presented on the phage particle on cultured cells, care must always be taken. The data can vary greatly depending on the number of cells remaining in each 96 well after ELISA, the display level of the individual phage antibodies, the background binding due to phage coat proteins binding to the ELISA well, etc. The signal intensities in the ELISA screenings were low. However, unlike ELISA against purified proteins, low signal in ELISA against cells is expected. This is primarily because unique or upregulated antigens in the targeted cells are not necessarily expressed at high levels, or at all, in the cells used for screening. Furthermore, some of the selected phage antibodies may bind the antigens with low affinity, regardless of the relevance of the antigens in question.
From our selection, the antibody BC5 was isolated, and its cognate antigen was identified as Mrps18a. Other MRPs such as MrpL11, MrpL12 and MrpL28 have been reported to be differentially expressed in tumour cells or tissue [
30,
31]. Knock-down of MRPL28 in pancreatic tumour cells resulted in decreased mitochondrial activity, increased glycolysis, and accelerated growth in vivo, a phenomenon commonly observed in cancer cells (the Warburg effect) [
31]. This shows that changes in the expression of Mrps affect the mitochondrial metabolism and thereby many of the parameters that control cell state (and, in turn, cancer progression).
Furthermore, it has been observed that some Mrps have functions in addition to being involved in mitochondrial translation. For example, Mrps29, another protein component of the small mitochondrial ribosomal subunit, binds GTP and may be involved in the control of apoptosis [
32,
33]. In addition, it has been reported that Mrps18b specifically binds to the viral transforming protein, EBNA-6, which targets Mrps18b to the nucleus in EBV-transformed lymphoblastoid cells. In the nucleus, Mrps18b binds retinoblastoma protein, which releases e2F1. Hence, Mrps18b may facilitate a lift of the RB-dependent block of S-phase entry [
34].
A specific role for Mrps18a in cancer has not been identified until now. Due to the low level of sequence similarity in the S18 protein family, functional conservation cannot be assumed.
At present, the focus of research is turning towards elucidating the role of mitochondria in cancer. Breast cancer markers and cancer markers in general are required for the further advancement of diagnostic and treatment strategies.
Acknowledgements
Benedicte Thuesen from Københavns Privathospital and the donors for providing normal breast biopsies. Seandean Harwood for proofreading the manuscript.