Keywords

Cancers and Cancer Progression

Cancers are diseases that arise through growth regulation defects but progress because cells lose cohesion with their tissue of origin, undergo architectural changes that allow them to invade surrounding tissues, and colonize distant organ systems. Broadly, cancers are classified as benign or malignant. Benign cancers grow locally without invading surrounding tissues and are only harmful if they press on and damage vital organs. In contrast, malignant cancers invade surrounding tissues and give rise to secondary tumors or metastases in organ systems that are removed from the primary tumor. Most human cancers arise in epithelial tissues, and these account for ~80 % of cancer-related deaths. In order to spread, these cancers must undergo very dramatic architectural changes in a process referred to as an epithelial to mesenchymal transition (EMT). The remaining tumors fall broadly into three groups: sarcomas arise in soft tissues of mesenchymal origin and represent only 1 % of human cancers; hematological malignancies arise from blood-forming tissues including cells of the immune system; finally neuroectodermal tumors include those cancers that arise in tissue derived from neuroectoderm and, whilst rare (~1.3 %), account for a disproportionate number of cancer-related deaths [1]. As alluded to above, cancers arise through a complex series of events, some of which occur through a sequence of somatic mutations and some through architectural remodeling that might arise through epigenetic processes. Of particular importance in architectural remodeling is altered expression of cytoskeleton proteins, cell adhesion molecules, and components of the extracellular matrix. Over the past 15 years the contribution of nuclear architectural proteins such as the nuclear lamins to this architectural remodeling has been investigated, but as yet no clear consensus on their contribution to cancer progression has emerged [2]. Historically, it makes every sense to assume that changes in nuclear architecture are central to cancer progression because historically the most common pathological feature of cancers is altered nuclear architecture. In this chapter I will summarize the historical context in which lamins and cancer collide and attempt to resolve the apparent contradictions in the underpinning literature.

Nuclear Morphology as Diagnostic and Prognostic Markers

For several decades now, nuclear morphology has been used to both diagnose malignancy and to predict patient outcomes [3]. Today the best known diagnostic test that uses nuclear morphology to detect malignancy is the cervical Papanicolaou (PAP) smear test [4]. In the test, enlarged and irregularly shaped nuclei are used for the initial diagnosis of cervical and uterine cancer. However, the same criteria can be applied to ovarian cancer, which, in combination with genetic changes, can be used to distinguish low- from high-grade cancers [5]. Clinical studies on breast cancer developed a semiquantitative scoring system based on three criteria: tubule formation within the tumor, nuclear pleomorphism, and mitotic counts. Of these three criteria nuclear pleomorphism included both quantitative and qualitative features, namely, nuclear size, regularity of shape, and uniformity of chromatin. In low-grade tumors (well differentiated) nuclei are small and round and have uniformly distributed chromatin. In contrast, in high-grade (poorly differentiated) tumors, nuclei vary in size, have “bizarre” shapes, and possess prominent and multiple nucleoli [6]. In breast cancer alteration in nuclear morphology might precede malignancy. Benign breast disease (BBD) is a condition that includes proliferative changes in breast tissue and is a risk factor associated with malignancy. In one control study it was reported that altered nuclear shape might identify those individuals with BBD that will go on to develop breast cancer [7]. Thus changes that lead to altered nuclear morphology might also promote malignancy. As a consequence, understanding which proteins support nuclear morphology may lead to better and more refined diagnostic and prognostic tools.

The Nuclear Matrix and Cancer

A number of studies have used sequential extraction of DNA, RNA, and soluble proteins to attempt to understand nuclear architecture and to identify proteins that support it. Early studies termed this structure the nuclear matrix or the nucleoskeleton, and at an ultrastructural level it appears as a network of filaments throughout the nucleus [8]. Whether or not this structure actually exists as opposed to being an artefact of the conditions of extraction has been questioned [9]. Nevertheless, comparisons of the biochemical composition of nuclear matrices between normal and malignant cells suggest that there are characteristic subsets of proteins in this compartment, which distinguish cancerous from normal cells and could be used as diagnostic or prognostic tools [10]. Most prominent amongst proteins that might form the nuclear matrix are non-myogenic nuclear actin, NuMA, nuclear lamins, LAP2a, and nuclear pore-associated proteins Nup153 topoisomerases I and II and Tpr [11]. Of these only the lamins have been systematically tested as potential diagnostic or prognostic tools, and so the rest of the chapter concentrates on studies relating to lamins.

Lamins and Lung Cancer

Lung cancers can be divided into two major disease types termed non-small-cell lung carcinoma (non-SCLC) and small-cell lung carcinoma (SCLC). Of these two types SCLC is poorly differentiated and has a markedly more rapid rate of progression. Since treating metastases of these two cancers involves completely different chemotherapy regimens, their correct diagnosis is critical [12]. In these cancers chromatin organization is the basis of diagnosis. At the level of light microscopy, chromatin is diffuse in SCLC whereas it is coarse and clumped in non-SCLC. At an ultrastructural level, this difference in appearance arises from a greater abundance of heterochromatin in the non-SCLC [13]. Two studies have attributed these chromatin structural differences to altered patterns of expression of nuclear lamins. By comparing levels of expression of lamins in non-SCLC and SCLC cell lines it was found that the expression of B-type lamins did not vary. In contrast, there was an 80 % reduction in the level of expression of lamins A/C in SCLC compared to non-SCLC lines. Furthermore, when SCLC cell lines were induced to differentiate into large (non-small)-cell carcinomas following transfection with the v-ras oncogene, expression of lamins A/C increased >10-fold whereas expression of a range of other nuclear proteins including B-type lamins, topoisomerases I and II, and B23/nucleophosmin did not change [14]. Different results were reported in a separate study. Similarly comparing lamin expression in lung carcinoma cell lines, B-type lamin expression was not noticeably different in those derived from SCLC and non-SCLC. In contrast, A-type lamins were expressed in all non-SCLC cell lines but were absent from 14 out of 16 SCLC cell lines—the opposite result from the first study. However, the patterns of expression were much more complex in cancer tissues. By staining of frozen sections of SCLC and non-SCLC, B-type lamins were expressed in all tumors, but in some non-SCLC (11 out of 23) a considerable proportion of tumor cells were negative for B-type lamins. In contrast, A-type lamins were absent or expressed at low levels in 14 out of 15 cases of SCLC but were expressed in all non-SCLC [15]. Neither of these studies attempted to understand whether these altered patterns of lamin expression contributed to the more rapid progression of SCLC. In retrospect, however, it is possible to understand the major diagnostic distinctions between the two cancer types (altered abundance of heterochromatin) in the light of lamin expression. A major feature of laminopathy disease is a loss of peripheral heterochromatin [16]. Thus in SCLC the dispersal and relative absence of heterochromatin may well arise from the lack of expression of A-type lamins.

Lamins and Breast and Ovarian Cancer

There is no particular reason to group breast and ovarian cancer (other than as examples of gender-specific cancers), and the grouping here is due to a paucity of studies as opposed to any other reason. Relating back to an earlier discussion, pathologists use nuclear morphology as a basis for diagnosis and in some cases prognosis in both diseases [5, 6]. In each diagnosis, nuclear shape is not considered in isolation but in combination with other disease features such as mitotic index and genetic defects such as aneuploidy [17]. Therefore it could be useful to understand if these two features are linked by a single determining factor. In a recent study, the expression of lamins was investigated in a very limited number of breast cancers and breast cancer cell lines. In breast cancer tissue, lamins A/C were either absent or aberrantly expressed in cancerous cells but were highly expressed in cells of surrounding normal tissue. In breast cancer cell lines, patterns of expression were more variable with some cell lines expressing normal levels of lamins A/C, whilst in other cell lines both A-type lamins were absent. In the same cell lines expression of lamin B1 and nuclear envelope transmembrane protein emerin were either unaffected or increased relative to normal breast epithelial cells [18]. In an attempt to link lamin expression to diagnostic features of the disease, lamin A/C expression was silenced in normal breast epithelial cells using shRNA. Silencing of lamin A/C expression led to irregularly shaped nuclei and aneuploidy, suggesting that loss of lamin A/C expression is linked to disease progression [18]. Using very similar approaches, lamin A/C expression and the consequences of their loss were investigated in ovarian cancer. In both ovarian cancer tissue and ovarian cancer cell lines lamin A/C expression was absent in 47 % of cases and heterogeneous in the remainder. Lamin A/C silencing in ovarian cancer cell lines led to an increase in nuclear volume and aneuploidy. Interestingly, in this instance those cells that became aneuploid appeared to undergo p53- and p21-induced growth arrest [19]. Thus, in this example loss of expression of lamin A/C is associated with diagnostic features of the disease but might retard rather than promote cancer progression.

Lamins and Hematological Malignancies

One of the problems in interpreting some of the studies referred to thus far is that findings were based on relatively small numbers of cell lines or cancer tissue samples and therefore lacked statistical power. More recent studies have looked at much larger numbers of patients and have attempted to link lamin expression to either disease-free survival or overall survival using Kaplan–Meier statistical tests. A very early study noted an absence of lamin A and C transcripts (but not B-type lamins) in neoplastic cells from patients with acute lymphoblastic leukaemia and non-Hodgkin’s lymphoma. Interestingly, lamin A and C transcripts could be detected in normal peripheral blood lymphocytes but only after mitogenic stimulation indicating that repression of LMNA, which codes for both lamin A and lamin C, was in some way related to the differentiation of lymphoid cells [20]. To understand the mechanism underlying lamin A/C repression and to investigate its link to survival, Agrelo and co-workers analyzed the promoter methylation status in human cancer cell lines from 17 tumor types as well as in primary leukaemias and lymphomas [21]. Importantly, the patient numbers were sufficient to guarantee statistical power in the study.

Although the cell lines used (70 in total) represented 17 different tumor types, only hematological malignancies were hypermethylated at LMNA promoters. Importantly, when LMNA CpG island-promoter methylation was found, it was directly linked to silencing. In primary cancers LMNA hypermethylation was detected in a minority (34 %) of nodal diffuse large B-cell lymphoma and Burkitt’s lymphoma (17 %) but was absent from cutaneous T-cell lymphoma, follicular lymphoma, and extranodal diffuse large B-cell lymphoma. In the case of nodal diffuse large B-cell lymphoma, LMNA promoter hypermethylation was associated with a statistically significant decrease in disease-free survival and overall survival, implying that LMNA silencing is a predictor of poor outcome only in this hematological cancer [21].

Lamins and Cancers of the Gastrointestinal Tract

Perhaps the best-studied cancers with respect to lamin expression are those of the gastrointestinal tract including the oesophagus, stomach, colon, and rectum. In an early study, immunohistochemistry was used to investigate the expression of lamins A/C and B1 in a variety of gastrointestinal cancers. On the whole, the results were quite variable although several key findings emerged. First, lamin A/C expression was reduced or absent from cancer cells in the majority of gastrointestinal cancers compared to normal epithelium and the protein was sometimes redistributed to the cytoplasm. Similar results were also found for lamin B1, although complete absence was less common, except in gastric adenocarcinoma. Whilst important, this study was mainly descriptive and did not attempt to link expression to cancer progression or survival [22]. In a more extensive study, lamin A/C expression was investigated in a large cohort (656) of colorectal cancer patients for whom survival data was available. Patients were found to either strongly express these lamins in cancer tissue (70 % of patients) or to display a complete absence of expression in cancer tissue (30 % of patients). Unexpectedly, patients that expressed lamin A/C in their tumor were twice as likely to die from cancer-related causes compared to patients without lamin A/C expression in their tumors (Cox’s hazard ratio = 1.85, p ≤ 0.005—[23]). In stark contrast to this report, an investigation of cancer tissue collected from 370 patients with stage II and III colon cancer showed an association of lamin A/C absence with disease recurrence and poor survival. However, poor prognosis was only found in patients with stage III colon cancer who had not received adjuvant chemotherapy (p ≤ 0.01—[24]). There were several important differences between the two studies that might account for these different results. Firstly, in the earlier study stage I adenocarcinomas were included as were rectal cancers and the study did not distinguish between patients who had and patients who had not received adjuvant chemotherapy [23]. If the differences are due to the inclusion of other regions of the intestinal tract, it implies that variation in lamin expression might reflect some aspect of gut architecture that differs depending upon location. This suggestion is reinforced by a third study on primary gastric carcinomas (GC). In this study 126 GC tissue samples were investigated for lamin A/C expression at both an mRNA and a protein level. Decreased levels of lamin A/C expression were observed in 56 % of patients, and this was correlated with poor differentiation (p ≤ 0.034) and poorer prognosis (p ≤ 0.034—[25]).

Lamins and Prostate Cancer

In a recent proteomic study, paired (benign and tumor) samples were collected from 23 low-grade and 26 high-grade tumors and subjected to two-dimensional differential gel electrophoresis (2D-DIGE). Of 19 abundant proteins that were identified by mass spectrometry as differentially expressed between tumor grades, lamin A was statistically highly discriminatory between low- and high-grade tumors (p ≤ 0.0003—[26]). In a follow-up study, the same group found that lamin A/C expression was concentrated at the invasive front of prostate cancer tissue. They went on to show that silencing of lamin A/C in prostate cancer cell lines inhibited cell growth, colony formation, migration, and invasion, whilst over-expression of lamin A/C stimulated the same processes [27]. In a complementary study Helfand and co-workers showed that instead of its normal uniform perinuclear distribution, prostate cancer cell lines are enriched for lamin B-deficient microdomains (LDMDs), which in turn over-express lamins A/C. In human prostate cancer tissue, the frequency of occurrence of LDMDs increased with tumor grade, again suggesting that increased expression of lamin A/C, in this case associated with lamin B1 deficiency, is correlated with tumor progression [28].

Lamins and Skin Cancers

Skin cancers are very heterogeneous and include sunlight-induced cancers such as melanoma, which can progress very rapidly; squamous cell carcinomas (SCCs), which have intermediate progression rates; and basal cell carcinomas (BCCs), which grow very slowly and rarely metastasize [29]. As their names suggest, the different skin cancers reflect their origins within the epidermis. Melanomas arise from melanocytes, BCCs from basal keratinocytes, and SCCs from the squamous layers of the skin. Studies on the expression of lamins in skin cancers also highlight why it might be that it has been hard to associate lamin expression with a consistent outcome in different and even in similar patient groups. In normal skin lamin A is expressed in suprabasal keratinocytes but is absent from basal keratinocytes. Lamin C is mostly absent from basal keratinocytes but can be observed in some basal cells. In contrast, lamins B1 and B2 are expressed in all epidermal layers although curiously lamin B1 is depleted in dermal fibroblasts [30]. In BCC lamin B1 and B2 expression was relatively constant whilst lamin A and lamin C varied considerably. In BCCs showing a high proliferative index, lamin A was typically absent or expressed at low levels. In contrast lamin A expression was higher in BCCs with a lower proliferation index but lamin C was absent or displaced from the nuclear periphery [30, 31]. In contrast, in SCCs lamin A was expressed at relatively high levels but lamin C was absent or displaced [31]. In order to explain these observations the authors suggested that the origins and progression of the cancers mapped to their lamin expression patterns in the different layers of a particular tissue. Since BCCs arise from basal cells which lack lamin A at an early stage they are lamin A negative and lamin C positive and have a relatively high proliferative index. As these cancers differentiate and their growth rates slow down, lamin A expression is up-regulated but lamin C is downregulated. Finally, SCCs that arise from squamous (lamin A positive) cells express lamin A, but this does not appear to impair proliferation rates [31]. What these two studies illustrate is a need to understand how lamin expression is linked to normal tissue organization before any strong links with tumor progression can be made or understood.

Mechanisms and Conclusions

At face value it is hard to reconcile some of the apparent contradictions in the current literature relating to lamins and cancers. Even within clinical studies that have statistical power, some point to downregulation of lamins A/C being associated with poor patient outcome [21, 24, 25] whilst others point to up-regulated expression of lamins as a prelude to metastatic disease [23, 27]. To some extent this illustrates the complexity of cancer biology and the different types of cancer being investigated. For example in B-cell lymphoma it is likely that expression of lamins A/C promotes differentiation [21], and therefore it might be expected that epigenetic changes that prevent this expression are likely to maintain B-cells in a primitive and more aggressive state. Cohort studies are also inherently problematic, and this is illustrated in the two large cohort studies on colorectal cancer. The Willis study utilized a cohort collected throughout the Netherlands dating back to 1986 [32]. Patients within the study group are thus unlikely to have received adjuvant chemotherapy. In contrast, the cohort specimens used in the Belt study were collected between 1996 and 2005 from patients undergoing surgical resection in the same hospital, some of whom were treated with adjuvant chemotherapy [24]. In the latter study, the follow-up period was 57 months, whereas in the Willis study the follow-up period was 84 months. Finally, the Belt study was limited to patients with colonic cancer, whereas the Willis study included patients with rectal cancer. Thus it is hard to compare two studies, which at face value appear to ask exactly the same question and arrive at directly contradictory results. Finally, the use of single biomarkers can be problematic, and, indeed, the studies of Tilli and co-workers suggest that subtle changes in the balance of lamin expression are usually observed, at least in skin cancers. Therefore studies including the expression of multiple lamin subtypes are definitely preferable.

The idea that lamin expression patterns in cancers are linked to where they arise harks back to a notion of how cancers arise. In skin cancers as has been described the different cancers are classified according to the cells from which they arose (basal keratinocytes, squamous cells, or melanocytes). Whilst a detailed study on lamin expression in melanoma has not been performed it does appear that for BCC and SCC the patterns of lamin expression do map quite nicely to the patterns observed in the basal cells and squamous cells. This might also be true of colorectal cancers. In the colon stem cells express lamin A but not lamin C. The transit amplifying cells do not express lamin A or lamin C, whilst the differentiated epithelial cells express both A-type lamins (Fig. 1—[23]). It is not absolutely clear where colonic adenocarcinomas arise, but there are two theories. The top-down theory supposes that the cancers arise in differentiated epithelial cells at the top of crypts and become more stem cell like. The bottom-up theory supposes that the cancers arise in the stem cell niche but somehow maintain a progenitor phenotype as they migrate upwards [1]. Either way it might be expected that the outcome of this switch to or maintenance of a progenitor phenotype is that lamin C expression is lost but lamin A expression is maintained. This again suggests that looking at individual lamin subtypes in the same study is important.

Fig. 1
figure 1

Lamin A/C expression in the colonic crypt. The different cell layers/stages in the colonic crypt as well as those of several other types of epithelia have different levels of lamin A/C expression. Cells at the base of the crypt are the precursor stem cells and the least differentiated, yet they express high levels of lamin A and no lamin C. As the cells differentiate and migrate up the crypt, lamin A expression is greatly reduced, and then as they approach the top of the crypt both lamins A and C become more and more expressed until there are very high levels of both in the most differentiated cells at the top of the crypt

Full size image

One area of convergence in the literature is the implication that expression of lamins in certain epithelial cancers might promote invasive behavior. In prostate cancer cell lines, over-expression of lamin A/C promotes cell survival, cell motility, and invasiveness, and this appears to be coordinated through the phosphadityl inositol-signaling pathway PI3K/AKT/PTEN [27]. In SW480 colorectal cancer cells expression of lamin A also promotes cell motility by remodeling the actin cytoskeleton as a result of up-regulation of actin-bundling proteins such as T- and F-plastin which leads to loss of expression of cell adhesion molecules such as E-cadherin, apparently promoting an epithelial transition [23, 33, 34]. Similar properties have been attributed to vimentin in breast cancer [35], which signals via the transcription factor slug and oncogenic Ras.

Thus future studies could be concentrated in two areas. Firstly, focusing on how lamins A/C as well as other intermediate filament proteins such as vimentin promote cell motility and invasiveness in several epithelial cancers might reveal convergent signaling pathways that could be influenced by an overall balance of the expression patterns of different lamin subtypes, as has been suggested in a recent study [36]. Secondly, a thorough understanding of how subtle changes in lamin expression influence pathways that promote either metastasis or survival of cancer cells or both, the design of quantitative methods to interrogate archival material from cancer patient cohorts, and careful consideration of factors that might skew data together could help to generate more certainty and consistency.