Background
Cell adhesion and migration are fundamental processes that occur during organogenesis, neural development, tissue regeneration and immune response, all of which require communication between cells and interaction of cells with their microenvironment. These two processes are also critical for tumor cells traveling to distant sites during metastasis [
1,
2]. Recent studies on molecules involved in cancer metastasis have found that several neural cell recognition molecules are abnormally expressed and functioning in clinical patient samples and in
in vitro tumor models [
3‐
5]. Surface proteins, such as N-CAM [
6,
7], Ng-CAM [
8], L1CAM [
9,
10] and neogenin [
11], which predominantly exert effects in nervous system development, have been demonstrated also to facilitate tumor cell progression in certain types of cancer.
L1CAM (CD171) is the initial member of the L1 family of immunoglobulin superfamily proteins and has pivotal roles in mediating the correct formation of neuronal connections during embryo neurogenesis [
12‐
14]. L1 and its homologous cell adhesion molecules are distributed mainly in the central and peripheral nervous systems. With six Ig-like and five fibronectin type III (FN III) domains in the extracellular region and a conserved intracellular cytoplasmic tail, this transmembrane glycoprotein possess sufficient functioning domains to interact with guiding cues or extracellular matrix proteins. Such interactions of neurons with their immediate environment instruct cell and/or axonal movement [
15]. L1CAM performs important functions in neuron-neuron adhesion, neuronal migration, neurite extension and fasciculation, axon outgrowth and synaptic plasticity [
5,
13,
15‐
17]. The importance of L1CAM in the nervous system can be emphasized by the severe syndromes that result from various L1 gene mutations categorized as L1-syndrome [
12,
13,
18‐
20]. One of the most severe results of this syndrome is hydrocephalus, which in many cases is due to mutations causing production of truncated L1 ectodomain, which is secreted. On the other hand, Kalus et al. [
21] found that L1-dependent neurite outgrowth requires highly regulated proteolysis of L1 at the cell surface. Otherwise the cellular microenvironment would be unfavorable for axon outgrowth. These findings shed some light on the potential role of L1 proteolytic cleavage and release of soluble L1 in facilitating neuron migration and axon growth cone protrusion in the nervous system.
Normally, non-neuronal expression of L1 can be found only in Schwann cells in the peripheral nervous system, in some lymphocytes and in part of the renal system [
22,
23]. But recent studies unveiled L1CAM's abnormal presence in glioma, melanoma, ovarian, colon and pancreatic cancers [
4,
10,
24,
25]. According to these studies, tumor cells tend to use the same mechanism involving L1 as neurons do in brain development to possess an increased migratory ability. Proteolysis of L1, however, is not tightly regulated as it is in the nervous system, and is constitutive. The abnormal expression of L1 in cancerous tissues compared to the normal tissues suggests its potential as a marker for tumorigenesis.
Soluble L1 resulting from ADAM10 cleavage has been found in serum and ascites fluid from patients with ovarian carcinoma or endometriosis [
26‐
28]. This proteolyzed form is proposed either to
cis-interact with integrin αvβ3, αvβ5 [
29] on the cell surface, or to become integrated into the immediate extracellular matrix [
30]. Studies from other research groups [
27,
31] and ours [
32] have shown that in colon cancer and glioma models, L1CAM expression and proteolysis can be detected at the invasive front of cell cultures, indicating the role of L1 shedding in the pioneering stage of cancer cell migration. Expression and cleavage of L1 in breast cancer cells can be detected both directly from the cell surface and from secretory vesicles [
9,
30]. Whether this soluble form can facilitate breast cancer cell migration or have any roles in other steps of metastasis is still unclear.
In view of previous studies, our current work aims at elucidating the function of L1CAM, especially its soluble form, in breast cancer cell migratory processes. From established datasets on clinical samples of breast cancer, L1 is found to be among the top ranking of genes whose expression is up-regulated, and expression is high in cells with active migration ability in vitro. Over expression of L1 ectodomain alone is adequate to increase cell adhesion to ECM and increase transmigration, and this effect can be blocked with anti-L1CAM antibody. We provide novel evidence to show the soluble form of L1 can be a chemotactic signal to direct cell migration. These results further emphasize the potential of L1CAM to be a target for breast cancer treatment.
Discussion
From a neural recognition molecule in brain development to a potential marker in cancer progression, this
de novo function of L1CAM has been uncovered in several cancer types [
4,
39]. We analyzed previously generated datasets of clinical samples and found that L1 expression levels were correlated with breast cancer stage progression and also were higher in more metastatic breast cancer cell lines that we used for our experiments here. We present new insights into how soluble L1 facilitates breast cancer cell motility by demonstrating that ectopic L1ED expression increased cell adhesion and migration in less migratory cell lines. In addition, we displayed a new function of shed L1 as an extracellular attractant in cancer cell migration.
Molecular profiling has been a great aid in sorting out potential markers for early cancer diagnosis. From previously established databases of breast cancer clinical samples or
in vitro cultured cell lines, the L1CAM gene or the region in the X chromosome where it is located (Xq28) [
40,
41] has been found to be abnormally amplified in some cases [
33,
42,
43]. Its over-expression also is correlated with progressing stages of cancer in patients (Figure
1A, [
33]). Results from analysis of our cultured cell lines revealed a similar trend (Figure
1B and
1C), with the less migratory MDA-MB-468 cells expressing the lowest level of L1CAM protein. Our laboratory also found L1 is expressed in invasive and metastatic SUM149 breast cancer cell line (data not shown) and others have shown L1 expression in MDA-MB-231 and -435 cells as well [
9]. Though the mechanism of how L1 is upregulated in metastatic tumors is still undefined, it is coincidently upregulated with ADAM10 as target genes of β-catenin/TCF signaling, with co-expression at the invasive front of colon carcinomas [
44]. Effects of various growth factors in cancer tissues [
45] also may contribute to the abnormal presence of L1CAM. Interestingly, analysis of 145 primary breast tumors and 51 breast cancer cell lines (including MDA-MB-231) in another study [
46] did not show that the expression of L1CAM was correlated with genome copy number, suggesting that expression was not due to genomic aberrations (e.g. amplification). However, they did not give L1 expression results for individual cell lines. We believe L1 promotes breast cancer progression by its upregulated expression and coincident proteolysis by ADAM10 to release the soluble L1ED, which then exerts its autocrine/paracrine stimulatory effect.
L1 functions in the developing nervous system mostly as a cell surface adhesion/recognition protein by homophilic binding and heterophilic interaction with ECM and other migration guiding cues [
15]. It can become endocytosed and can activate the intracellular MAP kinase pathway [
47‐
49] to elicit its effects. Regulated proteolysis has been detected in developing mouse brain [
9]. However, the presence of unregulated levels of soluble L1ED during nervous system development causes severe forms of L1 syndrome [
12,
13,
18‐
20], and constitutive L1 shedding can favor cell migration from tumors [
3,
28]. Gutwein and colleagues [
30] have reported that in the AR breast cancer cell line, L1 cleavage by ADAM10 can be detected to occur both in minute vesicles termed exosomes and at the cell surface to be released into the culture medium. These two forms of cleaved L1 both can be considered to be forms of soluble L1 (sL1). Our analysis corroborated their study on L1CAM localization in intracellular vesicles and cleavage at the membrane-proximal end in different breast cancer cell lines. Although Mechtersheimer et al. [
9] found only a slightly detectable band at around 30 kDa to suggest L1 proteolysis in MDA-MB-231 and MDA-MB-435 cells in their study, this difference could be a result of different antibodies being used or different culture conditions. Alternatively, the cytoplasmic domain already may have been degraded (presumably by γ-secretase), which sometimes occurred in our breast cancer cell lines as well (data not shown). Consistent with this idea is that they were able to clearly demonstrate soluble L1ED in the cell culture supernatant from those cell lines [
9].
sL1 thus generated has been reported in various cancers to be an active factor in angiogenesis [
38], anti-apoptosis [
50] and cell migration [
36]. The mechanisms proposed for sL1 working in tumors can be categorized based on to which part a receptor would primarily bind. With an RGD motif on its sixth Ig domain, the sL1 retains similar roles as cell-surface L1 and still can recruit and bind to integrins such as αvβ3, αvβ5, and αvβ3 [
9,
29] to activate FAK and Src at focal adhesion complexes. The signal transduction initiated thereafter can stimulate cytoskeleton protein reassembly to generate directional membrane protrusions. On the other hand, sL1 also can be integrated into the immediate extracellular matrix [
51] by binding to laminin or proteoglycan as a chemoattractant for any following cells to attach and adhere. Either way, sL1 provides flexible connections with the ECM for cells to mobilize [
52]. Since cell migration speed potentially depends on the turnover rates of adhesion and de-adhesion cycles [
53], fast and small focal complexes recruited by L1 binding and signaling actually facilitate the event of cell movement [
54]. Our results obtained by overexpressing sL1 in MDA-MB-468 cells supported such a mechanism (Figures
3 and
4). Specifically, ectopic L1ED expression alone is sufficient to cause more cells to adhere to matrix and then to migrate. However, no significant change was observed in the Matrigel™ cell invasion assay. This suggests that soluble L1 serves mostly as a released factor to allow adhesion and exploration of the cell's immediate environment. In order to invade through the ECM, gene expression changes causing proteolysis and degradation of the matrix are required [
54,
55], which evidently are not achieved by adding sL1 alone in a short time period.
Given that L1CAM is rarely present other than in the nervous system, immune effector cells, and kidney under normal circumstances [
23,
41], its constitutive expression and shedding in tumors makes it an ideal marker for cancer detection and treatment. For example, short hairpin RNA targeting L1CAM, which has been found to impair axon outgrowth [
56] in normal neurons, can disrupt cell proliferation and neurosphere formation of brain tumor cells [
57]. Our lab has found that attenuating L1CAM by this method in glioma cells decreased focal complex turnover, reduced cell motility
in vitro, and halted brain invasiveness
in vivo[
32]. Here, we showed that shRNA targeting L1 in breast cancer cells can weaken cell adhesion and transmigration ability to a significant extent. Also, antibodies blocking L1, which have been used by others [
24] and us [
32,
58] to inhibit cancer cell growth and motility, were shown to reduce breast cancer cell adhesion and migration in our assay. These results point to L1CAM as being an over-expressed cell surface molecule in tumors and that it contributes importantly to cell migratory behavior.
PMA is reported to elevate L1 shedding by activating the PKC pathway [
59]. In our experiments, PMA treatment increased sL1 in the culture medium, whereas the overall L1 protein expression level remained unchanged. This could be because activation of the MAPK network upon addition of PMA initiated gene expression [
60,
61], but did not increase L1CAM expression directly. Whereas, the expression level of ADAM10 or another protease involved in ADAM maturation [
62] may be promoted, so that more surface L1, together with other protein substrates, would be cleaved. For the same reason, more MDA-MB-468 cells migrated toward conditioned medium from PMA activated cells. In that analysis, it is hypothesized that soluble L1 in conditioned medium might have bound to the coating matrix like integrated tracks in the ECM, attracting the cells to migrate and/or L1 could have exerted its effects in its soluble form. Additionally, homophilically-bound L1 fragments might have caused integrin recruitment on cell protrusions to form focal complexes and speed up the overall motility [
53]. The reason why supernatant from PMA-activated cells is more attractive than supernatant from 468-L1ED cells could be because more active cell responses were initiated via the PKC pathway stimulated by PMA. Nonetheless, the dramatic increase in transmigration could be totally reversed by addition of anti-L1CAM antibody (Figure
6C), demonstrating that the attractant was the L1ED.
Methods
Cell culture conditions
Human breast cancer cell lines MDA-MB-231, MDA-MB-435 and MDA-MB-468 were used. MDA-MB-231 (ATCC No. HTB-26) was a gift from Dr. Ulhas Naik (Univ. of Delaware), MDA-MB-435 cell line was gifted by Dr. Danny Welch (Univ. of Alabama, Birmingham) and MDA-MB-468 cells were from Dr. Leslie Krueger (A.I. DuPont Hospital for Children). These three cell lines were maintained in DMEM (Mediatech Inc., Herndon, VA), supplemented with 10% bovine growth serum (BGS; Hyclone, Waltham, MA), 100 μg/mL penicillin/streptomycin (Mediatech Inc.) and 2 mM L-glutamine (Mediatech Inc.). Quail fibrosarcoma cell line QT6 cells were cultured in Medium 199 (Mediatech, Inc.) with 5% FBS, 20% tryptose phosphate broth, 2 mM L-glutamine and penicillin-streptomycin. Human glioma cell line U-87 MG was obtained and kept as described in [
36]. All cell lines were cultured at 37°C in a humidified atmosphere containing 5% CO2.
Antibodies and reagents
Three different antibodies against L1CAM were used. Polyclonal antibody NCAM-L1 (C-20) (sc-1508, Santa Cruz Biotechnology, Santa Cruz, CA) recognizes the C-terminal region of L1CAM protein, monoclonal antibody 5G3 (sc-33686, Santa Cruz Biotechnology) recognizes the first Ig domain and UJ127 recognizes the 5th fibronectin repeat (GTX23200; Gene Tex, Irvine, CA). Monoclonal antibody against β-tubulin (Developmental Studies Hybridoma Bank, Iowa City, Iowa) and mouse IgG whole molecule (Jackson ImmunoResearch, West Grove, PA) were used for controls where indicated. Phorbol 12-myristate 13-acetate (PMA) was purchased from Sigma (Sigma-Aldrich Corp., St. Louis, MO).
RT-PCR analysis
RNA extraction and generation of first strand cDNA was carried out using PureLink total RNA purification system (Invitrogen, Carlsbad, CA) and the SuperScript III First-Strand Synthesis System (Invitrogen). PCR was performed using the following primers with the length for each product indicated in parenthesis:
L1CAM [GeneBank: 3897] sense, 5'-TACCGCTTCCAGCTTCAG -3';
antisense, 5'- TGATGAAGCAGAGGATGAGC -3' (460bp)
actin [GeneBank: 60]: sense, 5'- GCTCGTCGTCGACAACGGCTC -3';
antisense, 5'- CAAACATGATCTGGGTCATCTTCTC-3' (353bp).
PCR was carried out as described [
32] using Master Mix (Promega, Madison, WI) in a thermal cycler (Techne, Burlington, NJ), and products were then electrophoresed on a 1.5% agarose gel and visualized by ethidium bromide staining.
Vector construction and lentivirus infection
L1 ectodomain (L1ED) fragment was generated from the pCDNA3-L1 vector kindly provided by Dr. Vance Lemmon (The Miami Project to Cure Paralysis, University of Miami, FL). The following primers with SpeI or XhoI (NEB, Ipswich, MA) cleavage sites were used to amplify the sequence: sense, 5'-GAAACTAGTCGCCGGGAAAG-3'; antisense, 5'- GCCTCGAGGAGGGAGCC- 3'. The resulting 3350bp L1ED was then inserted into a lentivirus vector (Lvv 1879; provided by Dr. John C. Kappes, Univ. of Alabama, Birmingham) containing a CMV promoter. Empty Lvv 1879 vector served as a negative control. Constructs obtained were both confirmed by DNA sequencing and used to make virus to infect MDA-MB-468 cells. MDA-MB-231 cells were infected with L1-shRNA vector or the mock control. An shRNA Lentiviral vector targeting human L1CAM (TRCN0000063917; cat No. RHS3979-97052304) and the non-targeting control vector pLKO.1 were purchased from Open Biosystem (Huntsville, AL). All four vector constructs were then transfected into HEK 293T/17 respectively, with the helper plasmid pMD.G and packaging plasmid pCMVΔR8.2 by ratio of 4:3:1 (20 μg: 15 μg: 5 μg for a 10 cm plate) using standard calcium phosphate method [
63]. Supernatants containing viruses were collected 48 and 72 hours after transfection and target cells were infected using 10 μg/mL polybrene. Stable infected cell lines with encoded vectors were selected by 2 μg/mL puromycin resistance and confirmed by immunoblotting or FACS analysis.
Western blot analysis
Protein extraction and western blot analysis was performed as described [
36]. In brief, plain MDA-MB-231, MDA-MB-435 and MDA-MB-468 cells or those stably infected cell lines were solubilized using RIPA buffer. Cell lysates were then quantified using the BCA Assay (Pierce Biotechnology, Pittsburgh, PA). Unless otherwise indicated, a total of 30 μg target proteins for each sample were probed with antibodies as indicated in each test.
Immunohistochemistry
Immunofluorescent staining was performed as before [
36] with minor differences. In brief, cells cultured on coverslips pre-coated with 200 μg/mL poly-L-ornithine (Sigma-Aldrich, St. Louis, MO) were directly stained with primary antibodies diluted in PBS with 5% heat-inactivated serum on ice for live staining. Alternatively, cells were fixed with 1% formaldehyde in PBS followed by primary antibody diluted in PBS with 5% normal goat serum and 0.03% Triton-X 100. In either case, the coverslips were then rinsed with PBS and incubated with Alexa Flour 594 secondary antibody (Molecular Probes, Invitrogen) diluted in PBS with 5% normal goat serum for 45 min-1 hr at room temperature. After washing, nuclei were stained in 10 μg/mL bisbenzimide (Sigma-Aldrich) and mounted. Thus prepared slides were visualized and digital images were taken by using a Zeiss LSM 510 Confocal imaging system (Carl Zeiss, Thornwood, NY, USA) with appropriate argon beam lasers.
FACS analysis
Cells were trypsinized, fixed (as described above) and stained with a saturating amount of mAb UJ127 or just secondary antibody Alexa Fluor-488 (Molecular Probes, Invitrogen) as control. Stained cells were examined and analyzed using a FACSCalibur flow cytometer (Becton Dickinson) using Cell Quest software.
L1 shedding analysis
L1 shedding analysis was performed as described [
9] with minor modifications. In brief, 10
6 cells were cultured in 35 mm tissue culture plates in complete medium for 24 hrs. The next day, culture plates were rinsed with PBS to remove growth factors and then 1 mL serum-free DMEM was replaced for an hour while cells remained normally attached. Where indicated, PMA (100 ng/mL) was co-incubated with the cells. After 1 hr incubation at 37°C, culture medium was separated from cell debris by centrifugation. The culture media were TCA precipitated and dissolved in LDS loading buffer (Invitrogen), while the pellets were lysed respectively as in western blot analysis. Proteins thus obtained were all subjected to western blotting for the presence of L1 with antibodies as indicated.
Cell adhesion assay
Cell adhesion assays were performed in 96-well flat bottom plates coated with different ECM proteins. Briefly, 50 μL/well solution of either fibronectin (10 μg/mL; Sigma-Aldrich) or laminin (10 μg/mL; Invitrogen) in PBS, or Matrigel™ (20 μg/mL; BD) in serum free DMEM were pre-added to 96-well plates and incubated at 4°C overnight. The plates were then rinsed and blocked with 0.2% BSA for 2 h at room temperature followed by three times PBS washing. Cells were then added to each well in triplicate and incubated for 45 min at 37°C. After washing, cells remaining attached to the plates were fixed and stained with a solution containing 0.5% crystal violet, 2% ethanol and 40% methanol in PBS. 100 μL SDS (1% wt/vol) was added to each well after washing, and the absorbance of the color substrate was measured with an FLUOstar OPTIMA microplate reader (BMG Labtechnologies, Offenburg Germany) at 595 nm. The percentage of bound cells was calculated by dividing the optical density of the adherent cells by that of the initial input cells, with BSA-coated wells subtracted as background reading. Where indicated, mAbs or mouse IgG negative controls were pre-incubated with the cells for 30 min at 4°C and then kept present during the adhesion period.
Cell transmigration and invasion assay
Cell transmigration assays were performed as described [
9] using 8.0 μm pore size Transwell inserts (Costar, Cambridge, MA) with some modifications. In brief, the bottom of the insert membrane was coated with either fibronectin or laminin at 10 μg/mL in PBS at 37°C for 90 min or 4°C overnight. Cells were serum-starved overnight (0.5% FBS), harvested with trypsin, washed and then resuspended in basal DMEM medium without serum. 10
5 cells in 0.1 mL were then added to the upper chamber of the inserts, and 0.6 mL DMEM with 10% serum was added to the lower chamber. Where indicated, conditioned medium obtained as described [
64] from cell culture supernatants was used instead in the lower chamber. After 12 h at 37°C, cells remaining on the upper side of the membrane were removed using cotton swabs, while the cells that migrated to the underside were fixed in methanol at room temperature for 30 min and then stained with crystal violet solution. The number of migrated cells was counted under a microscope in five fields at 100× magnification. Each assay was done in triplicate and presented as mean ± SEM. Cell invasion assays were done similarly to transmigration assays, only instead, transwell inserts were pre-coated with Matrigel™ at 1 μg/μL on the topside. Cells were allowed to invade for 24-48 hrs under normal culture conditions.
Statistical Methods
Data presented are mean ± SEM of at least three repeats. Student's t-test was used to analyze difference between two groups. ANOVA was used when more than two groups were involved, and then Student's t-test was further applied to analyze difference between groups. * or #, P < 0.05 was considered as significant; **, P < 0.01.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
YL and DSG designed experiments, conducted experiments and analyzed data. YL and DSG drafted and edited the manuscript. Both authors read and approved the final manuscript.