Access Type

Open Access Dissertation

Date of Award

January 2016

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Cancer Biology

First Advisor

Stephen P. Ethier

Second Advisor

Zengquan Yang

Abstract

Breast cancer is the most common cancer and the second leading

cause of cancer death in women. While ER-positive breast cancer subtypes are initially

well-managed by targeted therapies targeting estrogen signaling, many women are

suffering from recurrence of a more aggressive, hormone insensitive cancer 5 or more years after initial remission. Late recurrence of hormone resistant breast cancer in patients

who were previously successfully treated with anti-estrogen therapies worsens

overall long-term outcomes, and specific oncogenic mutations may be driving late recurring aggressive disease in these patients. More complete characterization of the oncogenome of

a tumor may allow for the possibility of customized therapy targeting each patient's

specific oncogene-activating mutations, and in doing so increase the probability

of durable remission over the long term. Basic research on cell lines modeling

different breast cancer subtypes makes possible discovery and characterization of novel

driving oncogenes and their context in a breast cancer subtype model.

The Wolf Hirschhorn Syndrome Candidate 1-Like 1 gene (WHSC1L1) is one of

approximately 50 genes in the chromosome 8p11-p12 amplicon, an amplified region

of the short arm of chromosome 8 found in 12-15% of human breast cancers, as well

as other cancer types such as lung. Amplification of the 8p11-p12 region is

most often found in breast cancers of the luminal B subtype. WHSC1L1 is a member of the NSD family of histone lysine methyltransferases, SET domain-containing proteins which

catalyze the addition of a methyl group to lysines on the amino-terminal tail

of histone H3 subunits. The WHSC1L1 gene expresses two known isoforms which code for two distinct proteins, WHSC1L1-long and WHSC1L1-short. The short isoform of WHSC1L1 codes for the first 647 of the 1437 amino acids present in the long isoform, and lacks the catalytic SET

domain and several PHD and PWWP chromatin interacting domains, containing a single

PWWP domain and a recently characterized acidic transactivation domain. In both

normal and tumor breast tissue, WHSC1L1-short is expressed at greater levels

than WHSC1L1-long. Several breast cancer cell lines established in the Ethier

lab harbor WHSC1L1 amplifications and also overexpress both isoforms of WHSC1L1.

The SUM44 cell line is a highly ER-positive cell line model of luminal B breast cancer isolated from a pleural effusion metastasis of a patient with aggressive disease. It is known that the

short isoform of WHSC1L1, WHSC1L1-short, is a potent driving oncogene in SUM44

cells, however the specific mechanism of WHSC1L1-short as an oncomodifier is not known.

To investigate WHSC1L1-short function as an oncogene in

SUM44 cells, we developed an shRNA knockdown model that specifically knocked down

expression of WHSC1L1-short through its unique 3' UTR sequence (shWHSC1L1-short), and a model that knocked down both WHSC1L1 isoforms (shWHSC1L1-total). We found that knockdown of both total WHSC1L1 and WHSC1L1-short alone negatively affected SUM44 proliferation, and that WHSC1L1-short knockdown had a larger effect than knockdown of both isoforms. After finding that WHSC1L1 expression was required for typical proliferation rates of SUM44 cells, we performed genome-wide expression profiling of SUM44 WHSC1L1-short and total WHSC1L1 knockdown cell lines relative to a control SUM44 line transduced with shRNA against lacZ. Again we found that knockdown of the WHSC1L1-short alone had a greater affect than total WHSC1L1 knockdown, this time on the number of significantly differentially expressed genes;

1131 genes were found to be differentially expressed in the WHSC1L1-short knockdown

cells relative to shLacZ control, while 238 genes were differentially expressed

in the total WHSC1L1 knockdown SUM44 cells relative to shLacZ control. Interestingly,

the ESR1 gene, which codes for the estrogen receptor alpha protein, was significantly

downregulated by WHSC1L1-short knockdown. This was confirmed by immunoblotting

with ERa antibody. While total WHSC1L1 knockdown also had a negative

effect on ERa protein levels in SUM44, knockdown of WHSC1L1-short alone

almost completely abrogated ERa in SUM44 as measured by western blot.

We subsequently found that SUM44 cells were extremely sensitive to treatment

with beta-estradiol, and that proliferation actually decreased upon as little

as 100 picomolar beta-estradiol treatment, with dose-dependent decreases in

proliferation as estrogen concentrations increased. We also found SUM44 cells to be

relatively insensitive to Tamoxifen. Knockdown of WHSC1L1-short reduced proliferation

of SUM44 cells in estrogen-free conditions, and treatment of SUM44 shWHSC1L1-short

cells with increasing concentrations of beta-estradiol resulted in a marginal increase in proliferation up to 100pM beta-estradiol, then proliferation decreased with increasing beta-estradiol concentrations similar to results seen in SUM44 shLacZ control cells.

After observing that WHSC1L1-short overexpression was required for expression

of ERa in SUM44 cells, we asked whether knockdown of WHSC1L1-short

affected genome-wide binding patterns of the estrogen receptor in SUM44 cells.

Interestingly we found that ERa was binding to thousands of genomic

loci in the absence of exogenous estrogen. Treatment with high doses (10nM) of beta-estradiol for 45 minutes resulted in an approximately even increase in ERa binding

across sites already bound in the absence of estrogen, with some additional weak

binding sites, but no significant changes in the pattern of ERa binding.

No ERa binding sites were detected in SUM44 shWHSC1L1-short cells under

estrogen-free conditions, and weak ERa binding was detected in SUM44

shWHSC1L1-short cells treated with 10nM beta-estradiol at loci where strong

ERa binding was observed in control SUM44 cells, suggesting that

WHSC1L1-short knockdown was reducing ERa expression levels, which

made less ERa available to bind to chromatin in SUM44 shWHSC1L1-short

cells.

Our investigation of WHSC1L1 oncogenic activity in SUM44 cells resulted in

the interesting observation that the short isoform of WHSC1L1 is required for

expression of the estrogen receptor alpha in these cells, and that ERa

is bound extensively to chromatin without activation of ERa by estrogen.

SUM44 is a model for luminal B breast cancer, and is highly ER-positive, and

expresses little to no progesterone receptor (PR). While the implications of

ERa expression dependence on WHSC1L1-short overexpression in SUM44

cells are not yet clear, the extensive binding of ERa to estrogen

response elements (ERE's) in the absence of exogenous estrogen and the

negative proliferative response of SUM44 to estrogen indicate that WHSC1L1 amplification

and overexpression may alter the biology of the estrogen receptor in breast cancers

harboring WHSC1L1 amplification and overexpression.

Additionally, the differences seen in ERa binding to chromatin and

the negative response of SUM44 cells to ERa agonists illustrate the

importance of researching ER-positive breast cancer using additional cell line

models rather than consistently using MCF7 cells to represent ER-positive disease.

The dominant role of the catalytically-inactive short isoform of WHSC1L1 in regulating

ERa expression and maintaining the proliferation rate of SUM44 cells

suggests that catalytically inactive isoforms of chromatin modifying enzymes

can be important regulators of gene expression. Interestingly, recent work

has shown that WHSC1L1-short is likely not regulating target gene expression

through histone methylation, but instead is acting as a co-factor for a different

chromatin-binding complex, the BRD4-CHD8 complex, which has been shown to be

recruited to superenhancer regions (marked by histone acetylation) by WHSC1L1-short,

which results in activation of pTEFb through BRD4 and directly activates target

gene transcription. It will be important to avoid assumptions about binding substrate

identities of catalytically inactive isoforms of future chromatin modifiers of interest, as

the catalytically inactive isoforms of these genes may also bind to chromatin substrates unrelated to the substrate of the catalytically

active isoform.

Included in

Oncology Commons

Share

COinS