Abstract
Background
Acantholytic squamous cell carcinoma (Acantholytic SCC) are epithelial tumors characterized by a loss of cell adhesion between neoplastic keratinocytes. The mechanism underlying loss of cell-cell adhesion in these tumors is not understood.
Methods
A retrospective analysis of acantholytic SCC (n=17) and conventional SCC (n=16, controls not showing acantholysis) was conducted using a set of desmosomal and adherens junction protein antibodies. Immunofluorescence microscopy was used to identify tumors with loss of adhesion protein expression.
Results
The vast majority of acantholytic SCC (89%) showed focal loss of at least one desmosomal cell adhesion protein. Most interestingly, 65% of these tumors lost expression of two or more desmosomal proteins.
Conclusions
Loss of cell adhesion in acantholytic SCC is most likely linked to the focal loss of desmosomal protein expression, thus providing potential mechanistic insight into the patho-mechanism underlying this malignancy.
Keywords: Acantholytic squamous cell carcinomas, Desmosomes, Desmocollin, Desmoplakin, Desmoglein
INTRODUCTION
Acantholytic squamous cell carcinoma (acantholytic SCC) is a histopathologic subtype of squamous cell carcinoma (SCC) characterized by focal loss of cell-cell adhesion. These tumors are typically well-differentiated. Further, acantholysis may be so prominent in focal areas that it produces a pseudoglandular pattern. In addition, dyskeratotic cells may be prominent (e.g. (1–4)). Approximately 2–4% of all SCC are classified as acantholytic SCC (5).
Acantholytic SCC most commonly occur in men on the sun-exposed skin of the face, the neck and the hands of older patients (3). Further, acantholysis has been postulated to be a risk factor for metastasis, specifically in patients with larger lesions (3, 4, 6). The mechanism underlying loss of cell-cell adhesion in these tumors is not understood; however, acantholysis is observed in patients and animals with impaired desmosome function (7, 8), suggesting the possibility that desmosomal defects might occur in acantholytic SCC.
Desmosomes are cell adhesion structures (junctions) that are abundant in epidermal cells. Structurally, these junctions can be divided into a transmembrane core, which is crucial for connecting the plasma membranes of neighboring cells, and a plaque that connects the transmembrane core to the keratin filaments of keratinocytes (see references in (7, 9)). The core is composed of transmembrane proteins that are sequence-related to cadherins, and which have been termed desmosomal cadherins (10). Two families of transmembrane proteins are found in desmosomes, desmocollins (DSC) and desmogleins (DSG). Both groups contain several proteins which are encoded by different genes (DSC1-3, DSG1-4). These proteins are differentially expressed in desmosome-forming tissues. It is assumed that the different desmosomal cadherins vary in their biological properties. For example, the complement of desmosomal cadherins expressed in a given cell might affect adhesive strengths between cells and the ability to migrate. The desmosomal plaque contains plakins; desmoplakin (DSP), plakophilin(s) (PKP) and junctional plakoglobin (JUP) (9). DSP plays a crucial role in connecting the desmosome to the keratin intermediate filament network (11). JUP is unique in that it integrates into two different junctions; desmosomes and adherens junctions (AJ). All of the proteins listed above have been shown or postulated to be crucial for cell-cell adhesion (7, 8, 11–13).
We assessed the expression of constitutive desmosomal proteins as well as the expression of two marker proteins of adherens junctions (E-cadherin and β-catenin) in acantholytic SCC and conventional SCC. Our results indicate that focal loss of desmosomal adhesion protein occurs in the majority (89%) of acantholytic SCC analyzed, providing a possible mechanistic explanation for the development of acantholysis in this tumor type.
MATERIAL AND METHODS
A searchable electronic database was used to retrieve cases of cutaneous acantholytic SCC or conventional SCC from 2010 to 2013 from the University of Colorado Dermatopathology Consultant archives. The cases were identified by typing in either “acantholytic squamous cell carcinoma” or “squamous cell carcinoma” in the search menu of the database. The archived slides were retrieved and re-examined by a dermatopathologist (JEF). Cases that were excluded from the study included those specimens that were superficial, lacked enough residual tissue on to the block to study or did not demonstrate acantholysis in at least 25% of the invasive portion of the tumor. Each specimen was assigned a database-generated identification number for patient confidentiality before being released for laboratory study.
A total of 17 acantholytic SCC and 18 SCC were identified that met the criteria required for inclusion in the study. All acantholytic SCC’s were classified as being moderate (2 cases) or well-differentiated (15 cases). Fourteen of the 17 cases were from males and three of 17 were from females. The mean age was 68.4 years (range was 53 to 89 years). This is consistent with previously published data which has reported that acantholytic SCC’s are more common on older male patients (3). The conventional SCC without acantholysis were classified as being poorly-differentiated (1 case), moderately-differentiated (2 cases) or well-differentiated (16 cases). Nine tumors were from males and nine tumors were from females. The mean age was 71.5 years (range was 53 to 89 years).
Antibody Staining
Formalin-fixed and paraffin-embedded tumor sections were stained for immunofluorescence microscopy with the following autobodies using standard protocols: DSC3 (U114; Cat #61093, Progen; Heidelberg, Germany), DSG1/2 (DG3.10; Cat #61002; Progen; Heidelberg, Germany), DSG3 (5G11; Cat #MCA2273T; AbD Serotec; Oxford, UK), DSP (2.15, 2.17; Cat #10R-D108A; Fitzgerald; Acton, MA), JUP (PG 5.1; Cat #10RP128A, Fitzgerald; Acton, MA), β-catenin (Cat #610153, BD Transduction Laboratories; San Jose, CA), E-Cadherin (24E10; Cat #3195, Cell Signaling; Danvers, MA), Keratin 5 (KRT5, Cat # PRB-160P, Covance; Emeryville, CA). For a description of the desmosomal antibodies used refer to the references (14, 15). Fluorescent dye-conjugated secondary antibodies were purchased from Invitrogen (Eugene, OR). Desmosomal and adherens junction protein antibodies were used in combination with KRT5 antibodies (double staining) to identify protein expression in epithelial tissues. Antibody dilutions were determined empirically using human skin as a substrate. Antigen retrieval was done as recommended by the manufacturer of the antibodies. Fluorescence staining was documented with a Nikon Eclipse 90I microscope equipped with a Coolsnap HQ2 and a DS-Fi1 camera. Image processing was done with the NIS Elements 3.10 imaging software (Nikon). All pictures were taken with the same lens (100X magnification) and each antibody was imaged with the same exposure time. Loss of fluorescent staining was confirmed by two independent observers.
RESULTS
Using acantholytic SCC (n=17) and conventional SCC (n=18, controls) tumor sections obtained from a clinical dermatopathology laboratory, we performed a retrospective study to determine whether junctional proteins are deregulated in these carcinomas.
Previous studies have not systematically analyzed the expression of constitutive desmosomal proteins in acantholytic SCC. The vast majority of acantholytic SCC (89%) in our study showed loss of at least one desmosomal protein (Table II). One acantholytic SCC showed normal expression of these proteins as judged by immunofluorescence microscopy (data not shown) while a second acantholytic SCC showed expression of all five markers, albeit with reduced expression levels of three desmosomal proteins (Table I and data not shown). Nevertheless, given that reduced expression is difficult to quantitate reliably by immunofluorescence microscopy, we restrict our discussion to samples in which proteins were apparently absent (see examples in Fig 1). Note that in most cases, loss occurred focally and did not extend throughout the entire tumor tissue, thus providing an internal control for our staining experiments.
Table II.
Loss of multiple junctional protein expression as determined by immunofluorescence microscopy
| Acantholytic SCC (n=17) | Conventional SCC (n=18) | |
|---|---|---|
| 1 Desmosomal protein lost | 4/17 (24%) | 3/18 (17%) |
| 2 Desmosomal proteins lost | 5/17 (29%) | 2/18 (11%) |
| 3 Desmosomal proteins lost | 2/17 (12%) | 0/18 (0%) |
| 4 Desmosomal proteins lost | 2/17 (12%) | 0/18 (0%) |
| 5 Desmosomal proteins lost | 2/17 (12%) | 0/18 (0%) |
| 1 AJ protein lost | 2/17 (12%) | 0/18 (0%) |
| 2 AJ proteins lost | 2/17 (12%) | 0/18 (0%) |
Table I.
Desmosomal protein expression in acantholytic SCC and conventional SCC (FL, Focal Loss)
| Tumor | DSG1/2 | DSP | DSG3 | DSC3 | JUP | E-Cadherin | β-Catenin |
|---|---|---|---|---|---|---|---|
| Acantholytic SCC 1 | |||||||
| Acantholytic SCC 2 | FL | FL | |||||
| Acantholytic SCC 3 | FL | FL | |||||
| Acantholytic SCC 4 | FL | ||||||
| Acantholytic SCC 5 | FL | FL | FL | ||||
| Acantholytic SCC 6 | |||||||
| Acantholytic SCC 7 | FL | FL | FL | FL | |||
| Acantholytic SCC 8 | FL | FL | |||||
| Acantholytic SCC 9 | FL | FL | FL | FL | FL | ||
| Acantholytic SCC 10 | FL | FL | |||||
| Acantholytic SCC 11 | FL | FL | FL | FL | FL | FL | FL |
| Acantholytic SCC 12 | FL | FL | |||||
| Acantholytic SCC 13 | FL | ||||||
| Acantholytic SCC 14 | FL | FL | |||||
| Acantholytic SCC 15 | FL | FL | FL | FL | FL | ||
| Acantholytic SCC 16 | FL | FL | |||||
| Acantholytic SCC 17 | FL | FL | FL | FL | |||
| Conventional SCC 1 | |||||||
| Conventional SCC 2s | |||||||
| Conventional SCC 3 | FL | ||||||
| Conventional SCC 4 | |||||||
| Conventional SCC 5 | FL | ||||||
| Conventional SCC 6 | |||||||
| Conventional SCC 7 | |||||||
| Conventional SCC 8 | FL | ||||||
| Conventional SCC 9 | |||||||
| Conventional SCC 10 | FL | FL | |||||
| Conventional SCC 11 | |||||||
| Conventional SCC 12 | |||||||
| Conventional SCC 13 | |||||||
| Conventional SCC 14 | |||||||
| Conventional SCC 15 | FL | FL | |||||
| Conventional SCC 16 | |||||||
| Conventional SCC 17 | |||||||
| Conventional SCC 18 |
Fig. 1.
Three acantholytic SCC (A–C) were stained with the antibodies indicated. In each of the three tumors shown, two desmosomal proteins show overlapping loss of synthesis in the same areas. Arrowheads point to areas in which the marker proteins are normally expressed. Arrows indicate focal loss of desmosomal protein expression. The open arrowhead in Fig 1A demarcates an area in which DSG3 staining appears to be cytoplasmic, suggesting functional inactivation of this protein. Desmosomal cadherins require cell membrane localization to function as cell adhesion proteins. (Final magnification, 100X)
The results listed in Table I and III show that in acantholytic SCC, DSG1/2 (note that the antibody used cross-reacts with two DSG isoforms, DSG1 and DSG2) and DSP were the proteins most likely to be lost, followed by DSG3 and DSC3. Further, our analysis showed that many acantholytic SCC showed loss of multiple desmosomal proteins (Table II). Acantholytic SCC9 showed loss of all desmosomal proteins assessed while maintaining expression of the adherens junction markers. Acantholytic SCC11 showed loss of all adhesion proteins tested, including the adherens junction markers. Strikingly, our conventional SCC controls showed far less loss of desmosomal protein expression than acantholytic SCC. Only 28% of these tumors lost at least one desmosomal marker. Further, from the 18 SCC assessed, none showed loss of more than 2 desmosomal markers (Table II). None of the AJ markers were lost in the SCC samples analyzed (Table II).
Table III.
Loss of junctional protein expression as determined by immunofluorescence microscopy. P values were determined using Fisher’s Exact Test. Note that p ≤ 0.05 is considered statistically significant.
| Acantholytic SCC (n=17) | Conventional SCC (n=18) | P-Values | |
|---|---|---|---|
| DSG1/2 | 12/17 (71%) | 2/18 (11%) | 0.000491 |
| DSG3 | 7/17 (41%) | 2/18 (11%) | 0.059908 |
| DSC3 | 5/17 (29%) | 1/18 (6%) | 0.087683 |
| DSP | 10/17 (59%) | 1/18 (6%) | 0.000945 |
| JUP | 4/17 (24%) | 1/18 (6%) | 0.177419 |
| E-Cadherin | 4/17 (24%) | 0/18 (0%) | 0.045455 |
| β-Catenin | 1/17 (12%) | 0/18 (0%) | 0.48517 |
| ≥1 Desmosomal protein lost | 15/17 (89%) | 5/18 (28%) | 0.000491 |
| ≥1 AJ protein lost | 4/17 (24%) | 0/18 (0%) | 0.045455 |
DISCUSSION
The patho-mechanism underlying acantholysis in SCC is not understood. Several studies have attempted to link defects in cell-cell adhesion in these neoplasias to the loss of specific proteins. Several groups identified E-Cadherin and syndecan-1 as proteins lost or functionally inactive in acantholytic SCC (1, 5, 16, 17). However, at least with respect to syndecan-1, it appears that its role in the disease remains unclear since one study did not confirm de-regulation of this protein in acantholytic SCC (16). Lastly, over-expression of IKKα has also been claimed to induce acantholytic SCC-like tumors in a mouse model (18). This is somewhat surprising given that IKKα plays an important role in the normal differentiation of epidermal keratinocytes. Nevertheless, it is possible that IKKα serves a different function in transformed cells when compared to normal keratinocytes. It remains to be seen whether IKKα de-regulation is a common occurrence in acantholytic SCC.
In the present study, we determined weather acantholytic SCC loose expression of adhesion proteins. Note that we define the term “lost” as the absence of antibody staining in immunofluorescence microscopy. Loss of an immunofluorescence signal is indicative of a significant reduction in protein synthesis or stability. We are aware that this finding is not identical to a null mutation, which completely abolishes protein synthesis. Nevertheless, the absence of protein detection in our system is likely to indicate severe defects in desmosome function.
Two of the desmosomal cadherins tested, DSG3 and DSC3, have previously been linked to acantholysis in the epidermis and the oral mucosa. We have demonstrated that loss-of-function mutations in the Dsg3 gene cause acantholysis in stratified epithelia of mice (19, 20). Similarly, using mouse models, we showed that epidermis-specific loss of Dcs3 leads to acantholysis and skin blistering (21), a phenomenon that appears to be replicated in patients with DSC3 gene mutations (7, 8). Nevertheless, DSG3 and DSC3 were not the primary targets of protein loss in acantholytic SCC. Further, our analysis failed to demonstrate statistical significance for our finding of reduced DSG3 and DSC3 expression (Table III). Instead, loss of DSG1/2 and DSP appeared to be statistically significant in our cohort of acantholytic SCC (Table III). Note that the antibody used to detect Dsg1/2 recognizes two desmogleins, DSG1 and DSG2, proteins with different expression patterns in the epidermis. DSG2 is only weakly expressed in the interfollicular epidermis (22). Further, this protein is down-regulated during keratinocyte differentiation while DSG1 is up-regulated, explaining the basal DSG2 and the suprabasal DSG1 expression in the epidermis (e.g. (22)). The biological consequences of DSG2 loss in the epidermis are not known since neither animal models nor human patients have been described with loss-of-function mutations in this gene. Given the low DSG2 expression levels in interfollicular epidermis, it is questionable whether this protein plays a major role in cell adhesion between epidermal keratinocytes. Loss of DSG1, on the other hand, has been linked to epidermal pathology (8). Recently, Samuelov and colleagues (12) described patients with a homozygous DSG1 mutation which is predicted to result in a loss-of-function phenotype. These patients develop spinous and granular acantholysis with split desmosomes in the epidermis.
The second most frequently lost desmosomal protein in our tumor group was DSP. Mouse studies have demonstrated that loss of DSP in the epidermis can lead to stress induced acantholysis (11). Interestingly, mice with a conditional DSP null mutation in the skin also formed less adherens junctions. In our acantholytic SCC group, only two tumors showed loss of both adherens junction markers examined (E-cadherin and β-catenin). In these tumors, DSP expression was lost as well.
Tanaka and colleagues recently reported patients with a genodermatoses caused by a DSP loss-of-function mutations which resulted in skin blistering at sites of exposure to mechanical stress, confirming that DSP is required for cell-cell adhesion in human skin as well (13).
The findings described above clearly demonstrate that all of the desmosomal proteins frequently lost in acantholytic SCC are essential to maintain tissue cohesion in the epidermis. Surprisingly, many acantholytic SCC tumors showed loss of several desmosomal proteins, with almost 12% of the tumors losing all five proteins examined (Table II). It is tempting to speculate that loss of multiple desmosomal proteins might facilitate acantholysis and potentially tumor progression and metastasis. Unfortunately, due to a lack of data regarding clinical outcome in our acantholytic SCC group, we currently cannot address this possibility. In our control SCC group, very few desmosomal proteins were lost. This further supports the hypothesis that loss of cell adhesion in acantholytic SCC is primarily due to impaired desmosome function, not a defect in AJ.
In summary, we predict that loss of cell-cell adhesion in acantholytic SCC is primarily a desmosomal defect with most tumors (82% in our study) showing loss of either DSG1/2, DSP or both proteins.
Acknowledgments
This work was supported by a grant from NIAMS/NIH to PJK (RO1 AR053892) and by a NIAMS/NIH grant to the University of Colorado School of Medicine Histology Core under award number P30 AR057212. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors wish to thank Dr. Maranke Koster and Jason Dinella (both UC-AMC) for critical reading of the manuscript. Special thanks to Dr. Michael Edwards (Director, Bioinformatics Core, Pulmonary Sciences, University of Colorado School of Medicine) and Xian Lu (Department of Statistics, Colorado School of Public Health) for conducting the statistical analysis shown in Table III.
Abbreviations
- DSG
Desmoglein
- DSC
Desmocollin
- DSP
Desmoplakin
- JUP
Junctional Plakoglobin
- SCC
Squamous Cell Carcinoma
References
- 1.Driemel O, Muller-Richter UD, Hakim SG, et al. Oral acantholytic squamous cell carcinoma shares clinical and histological features with angiosarcoma. Head & face medicine. 2008;4:17. doi: 10.1186/1746-160X-4-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kerawala CJ. Acantholytic squamous cell carcinoma of the oral cavity: a more aggressive entity? The British journal of oral & maxillofacial surgery. 2009;47(2):123. doi: 10.1016/j.bjoms.2008.04.018. [DOI] [PubMed] [Google Scholar]
- 3.Nappi O, Pettinato G, Wick MR. Adenoid (acantholytic) squamous cell carcinoma of the skin. Journal of cutaneous pathology. 1989;16(3):114. doi: 10.1111/j.1600-0560.1989.tb00024.x. [DOI] [PubMed] [Google Scholar]
- 4.Cassarino DS, Derienzo DP, Barr RJ. Cutaneous squamous cell carcinoma: a comprehensive clinicopathologic classification. Part one. Journal of cutaneous pathology. 2006;33(3):191. doi: 10.1111/j.0303-6987.2006.00516_1.x. [DOI] [PubMed] [Google Scholar]
- 5.Papadopoulou E, Tosios KI, Nikitakis N, Papadogeorgakis N, Sklavounou-Andrikopoulou A. Acantholytic squamous cell carcinoma of the gingiva: report of a case and review of the literature. Oral surgery, oral medicine, oral pathology, oral radiology, and endodontics. 2010;109(6):e67. doi: 10.1016/j.tripleo.2010.01.019. [DOI] [PubMed] [Google Scholar]
- 6.Cherpelis BS, Marcusen C, Lang PG. Prognostic factors for metastasis in squamous cell carcinoma of the skin. Dermatologic surgery: official publication for American Society for Dermatologic Surgery [et al] 2002;28(3):268. doi: 10.1046/j.1524-4725.2002.01169.x. [DOI] [PubMed] [Google Scholar]
- 7.Ganeshan R, Chen J, Koch PJ. Mouse models for blistering skin disorders. Dermatology research and practice. 2010;2010:584353. doi: 10.1155/2010/584353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Petrof G, Mellerio JE, McGrath JA. Desmosomal genodermatoses. The British journal of dermatology. 2012;166(1):36. doi: 10.1111/j.1365-2133.2011.10640.x. [DOI] [PubMed] [Google Scholar]
- 9.Neuber S, Muhmer M, Wratten D, Koch PJ, Moll R, Schmidt A. The desmosomal plaque proteins of the plakophilin family. Dermatology research and practice. 2010;2010:101452. doi: 10.1155/2010/101452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Koch PJ, Franke WW. Desmosomal cadherins: another growing multigene family of adhesion molecules. Current opinion in cell biology. 1994;6(5):682. doi: 10.1016/0955-0674(94)90094-9. [DOI] [PubMed] [Google Scholar]
- 11.Vasioukhin V, Bowers E, Bauer C, Degenstein L, Fuchs E. Desmoplakin is essential in epidermal sheet formation. Nature cell biology. 2001;3(12):1076. doi: 10.1038/ncb1201-1076. [DOI] [PubMed] [Google Scholar]
- 12.Samuelov L, Sarig O, Harmon RM, et al. Desmoglein 1 deficiency results in severe dermatitis, multiple allergies and metabolic wasting. Nature genetics. 2013;45(10):1244. doi: 10.1038/ng.2739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Tanaka A, Lai-Cheong JE, Cafe ME, et al. Novel truncating mutations in PKP1 and DSP cause similar skin phenotypes in two Brazilian families. The British journal of dermatology. 2009;160(3):692. doi: 10.1111/j.1365-2133.2008.08900.x. [DOI] [PubMed] [Google Scholar]
- 14.Kurzen H, Moll I, Moll R, et al. Compositionally different desmosomes in the various compartments of the human hair follicle. Differentiation; research in biological diversity. 1998;63(5):295. doi: 10.1046/j.1432-0436.1998.6350295.x. [DOI] [PubMed] [Google Scholar]
- 15.Kurzen H, Munzing I, Hartschuh W. Expression of desmosomal proteins in squamous cell carcinomas of the skin. Journal of cutaneous pathology. 2003;30(10):621. doi: 10.1034/j.1600-0560.2003.00122.x. [DOI] [PubMed] [Google Scholar]
- 16.Griffin JR, Wriston CC, Peters MS, Lehman JS. Decreased expression of intercellular adhesion molecules in acantholytic squamous cell carcinoma compared with invasive well-differentiated squamous cell carcinoma of the skin. American journal of clinical pathology. 2013;139(4):442. doi: 10.1309/AJCPTN4BNJYIRUWO. [DOI] [PubMed] [Google Scholar]
- 17.Bayer-Garner IB, Smoller BR. The expression of syndecan-1 is preferentially reduced compared with that of E-cadherin in acantholytic squamous cell carcinoma. Journal of cutaneous pathology. 2001;28(2):83. doi: 10.1034/j.1600-0560.2001.280204.x. [DOI] [PubMed] [Google Scholar]
- 18.Moreno-Maldonado R, Ramirez A, Navarro M, et al. IKKalpha enhances human keratinocyte differentiation and determines the histological variant of epidermal squamous cell carcinomas. Cell cycle. 2008;7(13):2021. doi: 10.4161/cc.7.13.6147. [DOI] [PubMed] [Google Scholar]
- 19.Koch PJ, Mahoney MG, Cotsarelis G, Rothenberger K, Lavker RM, Stanley JR. Desmoglein 3 anchors telogen hair in the follicle. Journal of cell science. 1998;111 (Pt 17):2529. doi: 10.1242/jcs.111.17.2529. [DOI] [PubMed] [Google Scholar]
- 20.Koch PJ, Mahoney MG, Ishikawa H, et al. Targeted disruption of the pemphigus vulgaris antigen (desmoglein 3) gene in mice causes loss of keratinocyte cell adhesion with a phenotype similar to pemphigus vulgaris. The Journal of cell biology. 1997;137(5):1091. doi: 10.1083/jcb.137.5.1091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Chen J, Den Z, Koch PJ. Loss of desmocollin 3 in mice leads to epidermal blistering. Journal of cell science. 2008;121(Pt 17):2844. doi: 10.1242/jcs.031518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Brennan D, Mahoney MG. Increased expression of Dsg2 in malignant skin carcinomas: A tissue-microarray based study. Cell adhesion & migration. 2009;3(2):148. doi: 10.4161/cam.3.2.7539. [DOI] [PMC free article] [PubMed] [Google Scholar]