Summary. We propose the elastic net, a new regularization and variable selection method. Real world data and a simulation study show that the elastic net often outperforms the lasso, while enjoying a similar sparsity of representation. In addition, the elastic net encourages a grouping effect, where strongly correlated predictors tend to be in or out of the model together.The elastic net is particularly useful when the number of predictors (p) is much bigger than the number of observations (n). By contrast, the lasso is not a very satisfactory variable selection method in the

/pdf/regularization-and-variable-selection-via-the-elastic-net-4z8q6j5o0i.pdf

Regularization and variable selection via the elastic net

Breast cancer patients with the same stage of disease can have markedly different treatment responses and overall outcome. The strongest predictors for metastases (for example, lymph node status and histological grade) fail to classify accurately breast tumours according to their clinical behaviour. Chemotherapy or hormonal therapy reduces the risk of distant metastases by approximately one-third; however, 70-80% of patients receiving this treatment would have survived without it. None of the signatures of breast cancer gene expression reported to date allow for patient-tailored therapy strategies. Here we used DNA microarray analysis on primary breast tumours of 117 young patients, and applied supervised classification to identify a gene expression signature strongly predictive of a short interval to distant metastases ('poor prognosis' signature) in patients without tumour cells in local lymph nodes at diagnosis (lymph node negative). In addition, we established a signature that identifies tumours of BRCA1 carriers. The poor prognosis signature consists of genes regulating cell cycle, invasion, metastasis and angiogenesis. This gene expression profile will outperform all currently used clinical parameters in predicting disease outcome. Our findings provide a strategy to select patients who would benefit from adjuvant therapy.

/pdf/gene-expression-profiling-predicts-clinical-outcome-of-5bu8dmnf6h.pdf

Gene expression profiling predicts clinical outcome of breast cancer

Mitogen-dependent progression through the first gap phase (G1) and initiation of DNA synthesis (S phase) during the mammalian cell division cycle are cooperatively regulated by several classes of cyclin-dependent kinases (CDKs) whose activities are in turn constrained by CDK inhibitors (CKIs). CKIs that govern these events have been assigned to one of two families based on their structures and CDK targets. The first class includes the INK4 proteins (inhibitors of CDK4), so named for their ability to specifically inhibit the catalytic subunits of CDK4 and CDK6. Four such proteins [p16 (Serrano et al. 1993), p15 (Hannon and Beach 1994), p18 (Guan et al. 1994; Hirai et al. 1995), and p19 (Chan et al. 1995; Hirai et al. 1995)] are composed of multiple ankyrin repeats and bind only to CDK4 and CDK6 but not to other CDKs or to D-type cyclins. The INK4 proteins can be contrasted with more broadly acting inhibitors of the Cip/Kip family whose actions affect the activities of cyclin D-, E-, and A-dependent kinases. The latter class includes p21 (Gu et al. 1993; Harper et al. 1993; El-Deiry et al. 1993; Xiong et al. 1993a; Dulic et al. 1994; Noda et al. 1994), p27 (Polyak et al. 1994a,b; Toyoshima and Hunter 1994), and p57 (Lee et al. 1995; Matsuoka et al. 1995), all of which contain characteristic motifs within their amino-terminal moieties that enable them to bind both to cyclin and CDK subunits (Chen et al. 1995, 1996; Nakanishi et al. 1995; Warbrick et al. 1995; Lin et al. 1996; Russo et al. 1996). Based largely on in vitro experiments and in vivo overexpression studies, CKIs of the Cip/Kip family were initially thought to interfere with the activities of cyclin D-, E-, and A-dependent kinases. More recent work has altered this view and revealed that although the Cip/Kip proteins are potent inhibitors of cyclin Eand A-dependent CDK2, they act as positive regulators of cyclin Ddependent kinases. This challenges previous assumptions about how the G1/S transition of the mammalian cell cycle is governed, helps explain some enigmatic features of cell cycle control that also involve the functions of the retinoblastoma protein (Rb) and the INK4 proteins, and changes our thinking about how either p16 loss or overexpression of cyclin D-dependent kinases contribute to cancer. Here we focus on the biochemical interactions that occur between CKIs and cyclin Dand E-dependent kinases in cultured mammalian cells, emphasizing the manner by which different positive and negative regulators of the cell division cycle cooperate to govern the G1-to-S transition. To gain a more comprehensive understanding of the biology of CDK inhibitors, readers are encouraged to refer to a rapidly emerging but already extensive literature (for review, see Elledge and Harper 1994; Sherr and Roberts 1995; Chellappan et al. 1998; Hengst and Reed 1998a; Kiyokawa and Koff 1998; Nakayama 1998; Ruas and Peters 1998).

/pdf/cdk-inhibitors-positive-and-negative-regulators-of-g1-phase-4awlfdse3n.pdf

CDK inhibitors: positive and negative regulators of G1-phase progression

Background A more accurate means of prognostication in breast cancer will improve the selection of patients for adjuvant systemic therapy. Methods Using microarray analysis to evaluate our previously established 70-gene prognosis profile, we classified a series of 295 consecutive patients with primary breast carcinomas as having a gene-expression signature associated with either a poor prognosis or a good prognosis. All patients had stage I or II breast cancer and were younger than 53 years old; 151 had lymph-node–negative disease, and 144 had lymph-node–positive disease. We evaluated the predictive power of the prognosis profile using univariable and multivariable statistical analyses. Results Among the 295 patients, 180 had a poor-prognosis signature and 115 had a good-prognosis signature, and the mean (±SE) overall 10-year survival rates were 54.6±4.4 percent and 94.5±2.6 percent, respectively. At 10 years, the probability of remaining free of distant metastases was 50.6±4.5 percent in the group with a...

/pdf/a-gene-expression-signature-as-a-predictor-of-survival-in-1f6irz2ok6.pdf

A Gene-Expression Signature as a Predictor of Survival in Breast Cancer

A catalogue of molecular aberrations that cause ovarian cancer is critical for developing and deploying therapies that will improve patients' lives. The Cancer Genome Atlas project has analysed messenger RNA expression, microRNA expression, promoter methylation and DNA copy number in 489 high-grade serous ovarian adenocarcinomas and the DNA sequences of exons from coding genes in 316 of these tumours. Here we report that high-grade serous ovarian cancer is characterized by TP53 mutations in almost all tumours (96%); low prevalence but statistically recurrent somatic mutations in nine further genes including NF1, BRCA1, BRCA2, RB1 and CDK12; 113 significant focal DNA copy number aberrations; and promoter methylation events involving 168 genes. Analyses delineated four ovarian cancer transcriptional subtypes, three microRNA subtypes, four promoter methylation subtypes and a transcriptional signature associated with survival duration, and shed new light on the impact that tumours with BRCA1/2 (BRCA1 or BRCA2) and CCNE1 aberrations have on survival. Pathway analyses suggested that homologous recombination is defective in about half of the tumours analysed, and that NOTCH and FOXM1 signalling are involved in serous ovarian cancer pathophysiology.

/pdf/integrated-genomic-analyses-of-ovarian-carcinoma-hbamiwnq4e.pdf

Integrated genomic analyses of ovarian carcinoma

We examined the role of promoter architecture, as well as that of the DNA-bending capacity of the E2F transcription factor family, in the activation of transcription. DNA phasing analysis revealed that a consensus E2F site in the E2F1 promoter possesses an inherent bend with a net magnitude of 40 +/-2 degrees and with an orientation toward the major groove relative to the center of the E2F site. The inherent DNA bend is reversed upon binding of E2F, generating a net bend with a magnitude of 25 +/- 3 degrees oriented toward the minor groove relative to the center of the E2F site. We also found that three members of the E2F family, in conjunction with the DP1 protein, bend the DNA toward the minor groove, suggesting that DNA bending is a characteristic of the entire E2F family. The Rb-E2F complex, on the other hand, does not reverse the intrinsic DNA bend. Analysis of a series of E2F1 deletion mutants defined E2F1 sequences which are not required for DNA binding but are necessary for the DNA-bending capacity of E2F. An internal region of E2F1, previously termed the marked box, which is highly homologous among E2F family members, was particularly important in DNA bending. We also found that a bent DNA structure can be a contributory component in the activation of the E2F1 promoter but is not critical in the repression of that promoter in quiescent cells. This finding suggests that E2F exhibits characteristics typical of modular transcription factors, with independent DNA-binding and transcriptional activation functions, but also has features of architectural factors that alter DNA structure.

A role for a bent DNA structure in E2F-mediated transcription activation.

Alternative processing of the immunoglobulin mu primary transcript results in regulated production of mRNAs encoding the secreted (microseconds) and membrane-bound (micro m) form of IgM heavy chain during B-cell development. To elucidate the basis for this control, we analyzed the expression of altered forms of the mu transcription unit. Deletion of intron sequence between the microseconds and micro m exons, which reduces the distance between the two poly(A) sites as well as the distance between micro m splice sites, enhances production of micro m RNA. Correct expression is restored by insertion of heterologous sequences, demonstrating that spacing is indeed the critical aspect. The altered spacing appears to affect poly(A) site usage rather than splice site usage, since it was the distance between the poly(A) sites rather than the distance between splice sites that was found to be decisive. Finally, removal of either the C mu 4 splice donor or the m1 splice acceptor, thus eliminating normal micro m splicing, does not increase usage of the microseconds poly(A) site. We therefore conclude that the major factor in determining the ratio of microseconds to micro m is a poly(A) site choice rather than a splicing choice.

Poly(A) site choice rather than splice site choice governs the regulated production of IgM heavy-chain RNAs

The control of cell proliferation is of central importance to the proper development of a multicellular organism, the homeostatic maintenance of tissues, and the ability of certain cell types to respond appropriately to environmental cues. Disruption of normal cell growth control underlies many pathological conditions, including endothelial proliferative disorders in cardiovascular disease as well as the development of malignant tumors. Particularly critical for the control of cell growth is the pathway involving the G1 cyclin-dependent kinases that regulate the Rb family of proteins, which in turn control E2F transcription factor activity. Because E2F is critical for regulation of cell proliferation, we sought to identify and to develop specific inhibitors of E2F function that might also be useful in the control of cellular proliferation. Moreover, because the control of E2F activity appears to be the end result of G1 regulatory cascades, the ability to inhibit E2F may be particularly effective in impeding a wide variety of proliferative events. We have used in vitro selection to isolate several unique RNA species from high complexity RNA libraries that avidly bind to the E2F family of proteins. These RNAs also inhibit the DNA binding capacity of the E2F proteins. We also show that an E2F RNA ligand can block the induction of S phase in quiescent cells stimulated by serum addition. As such, these data demonstrate the critical role for E2F activity in cell proliferation and suggest that such RNA molecules may be effective as therapeutic entities to control cellular proliferation.

Inhibition of cell proliferation by an RNA ligand that selectively blocks E2F function.

Previous experiments have demonstrated that adenovirus infection of human and mouse cells leads to an E1A-dependent activation of the DNA-binding capacity of a cellular transcription factor termed E2F. E2F binds to two sites in the adenovirus E2 early promoter which have been shown to be critical for E1A-dependent E2 early transcription, and the E2F-binding sites can confer E1A-induced transcription to a heterologous promoter. In addition, under a variety of circumstances, the increase in E2F-binding activity coincides with the activation of E2 transcription. We now find that, in addition to the E1A gene, another early viral gene, the E4 gene, is necessary for the activation of E2F-binding activity. Extracts prepared from human 293 cells, which express the E1A and E1B genes, had low levels of E2F activity, whereas infection of 293 cells with the E1A mutant dl312 increased E2F activity. This increase did not occur when 293 cells were infected with dl366, an E4 deletion mutant, nor was there an increase in E2F activity in HeLa cells infected with either dl366 or dl312; however, a coinfection with the two mutants yielded the normal wild-type increase in E2F. Furthermore, infection of HeLa cells with a high multiplicity of dl312, conditions that allow E4 gene expression in the absence of E1A, did not yield an increase in E2F activity. Thus, it appears that both the E1A gene and the E4 gene are directly involved in E2F activation. Measurements of E2 RNA production in a dl366 infection as compared with a wild-type or dl312 infection demonstrate that the E4 gene is essential for full E2 transcription. Furthermore, transfection assays of the E2 promoter demonstrate that, although E1A alone can trans-activate the E2 promoter, it is not as effective as the combination of E1A and E4 in the induction of the E2 promoter. We therefore conclude that the activation of the E2F factor leading to the activation of E2 transcription requires the combined action of both the E1A 289-amino-acid protein and an E4 product.

The adenovirus E4 gene, in addition to the E1A gene, is important for trans-activation of E2 transcription and for E2F activation.

The E2F family of transcription factors provides essential activities for coordinating the control of cellular proliferation and cell fate. Both E2F1 and E2F3 proteins have been shown to be particularly important for cell proliferation, whereas the E2F1 protein has the capacity to promote apoptosis. To explore the basis for this specificity of function, we used DNA microarray analysis to probe for the distinctions in the two E2F activities. Gene expression profiles that distinguish either E2F1- or E2F3-expressing cells from quiescent cells are enriched in genes encoding cell cycle and DNA replication activities, consistent with many past studies. E2F1 profile is also enriched in genes known to function in apoptosis. We also identified patterns of gene expression that specifically differentiate the activity of E2F1 and E2F3; this profile is enriched in genes known to function in mitosis. The specificity of E2F function has been attributed to protein interactions mediated by the marked box domain, and we now show that chimeric E2F proteins generate expression signatures that reflect the origin of the marked box, thus linking the biochemical mechanism for specificity of function with specificity of gene activation.

https://www.pnas.org/content/pnas/102/44/15948.full.pdf

Joseph R. Nevins

Papers

A role for a bent DNA structure in E2F-mediated transcription activation.

Poly(A) site choice rather than splice site choice governs the regulated production of IgM heavy-chain RNAs

Inhibition of cell proliferation by an RNA ligand that selectively blocks E2F function.

The adenovirus E4 gene, in addition to the E1A gene, is important for trans-activation of E2 transcription and for E2F activation.

Distinctions in the specificity of E2F function revealed by gene expression signatures