Muhammad Jalal Dr. Hoopes Genetic Regulation Seminar 4/17/2015 Epigenetic Features and Treatment of Acute Myeloid Leukemia I. Introduction The world of eukaryotic genetic regulation is exceedingly complex and multifaceted, with an incredible diversity of pathways, proteins, genes, and signals to regulate the fundamental on-off switch of transcription. In this paper, epigenetic regulations of acute myeloid leukemia will be considered through an in-depth analysis of five primary literature articles. Despite such a short scope of one of the many diseases that result from improper genetic regulation pathways, many themes which are commonly seen throughout genetic regulation will be studied extensively here. The proliferation of cancer over recent years has posed a call to extensive research on characterizing and targeting cancer related molecules and pathways. Understanding epigenetics, or mechanisms for transcriptional control which lie outside of that coded within the actual genome, has become increasingly important in molecular biology and cancer research. Besides genetic mutations, many types of cancers have been linked to epigenetic regulation. In this paper, three epigenetic modifications will be considered: DNA methylation of CpG sequences in the genome, changes to histone modification, and microRNA inhibition of tumor suppressors. Within these broad groups are further classifications; histone modification includes acetylation, methylation, deacetylation, demethylation, and others. The focus will be for the most part on acute myeloid leukemia (AML), the most common adult form of leukemia. The striking patterns of epigenetic regulation in such a small subset of all types of cancer will illuminate the complexity of epigenetics and add consideration for therapeutic targeting of these modifications.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Muhammad Jalal
Dr. Hoopes
Genetic Regulation Seminar
4/17/2015
Epigenetic Features and Treatment of Acute Myeloid Leukemia
I. Introduction
The world of eukaryotic genetic regulation is exceedingly complex and multifaceted,
with an incredible diversity of pathways, proteins, genes, and signals to regulate the fundamental
on-off switch of transcription. In this paper, epigenetic regulations of acute myeloid leukemia
will be considered through an in-depth analysis of five primary literature articles. Despite such a
short scope of one of the many diseases that result from improper genetic regulation pathways,
many themes which are commonly seen throughout genetic regulation will be studied
extensively here.
The proliferation of cancer over recent years has posed a call to extensive research on
characterizing and targeting cancer related molecules and pathways. Understanding epigenetics,
or mechanisms for transcriptional control which lie outside of that coded within the actual
genome, has become increasingly important in molecular biology and cancer research. Besides
genetic mutations, many types of cancers have been linked to epigenetic regulation. In this paper,
three epigenetic modifications will be considered: DNA methylation of CpG sequences in the
genome, changes to histone modification, and microRNA inhibition of tumor suppressors.
Within these broad groups are further classifications; histone modification includes acetylation,
methylation, deacetylation, demethylation, and others. The focus will be for the most part on
acute myeloid leukemia (AML), the most common adult form of leukemia. The striking patterns
of epigenetic regulation in such a small subset of all types of cancer will illuminate the
complexity of epigenetics and add consideration for therapeutic targeting of these modifications.
AML is a subset of leukemia, a group of cancers which affect the development of bone
marrow derived cells such as white blood cells and platelets. The term myeloid refers to a
particular lineage of hematopoiesis, the production of a variety of blood cells. A multipotent
hematopoietic stem cell (MHSC) can mature into either a myeloid progenitor cell or a lymphoid
progenitor cell; these cells can mature into more specialized cells, such as T-cells, macrophages,
and platelets. In AML, either the myeloid progenitor cell, known as a myeloblast, or a myeloid
stem cell, is unable to differentiate and as such accumulates in the bone marrow. Because these
undifferentiated cells have self-renewing potential, tumors can result; more dangerously, levels
of myeloid derived blood cells can decrease dramatically. Understanding the mechanism through
which AML tumor cells can be made to differentiate is of importance as a therapeutic option.
Some treatments, such as inhibitors of demethylases and deacetylases, have proven to be
extremely valuable in targeting these epigenetic differences and can provide a way to target
cancer cells without affecting normal cells. Two treatments, vorinostat and tranylcypromine, will
be considered in the primary literature.
The concepts that are illustrated in these papers and have also been shown in the class
include CpG Islands, LSD1, acetylation, methylation, microRNA, hematopoietic stem cells,
progenitor cells, and histone modification. Techniques seen in class such as western blotting,
bisulfite genomic analysis, immunofluorescence, a variety of assays, and heat maps of
microarray data will be seen in these papers. Furthermore, while outside of the scope of this
class, double stranded breaks and DNA repair mechanisms are commonly studied in molecular
biology and will also be studied in the histone deacetylase paper.
II. Concurrent DNA hypermethylation of multiple genes in acute myeloid leukemia (Melki
et al. 1999)
Melki et al. (1999) sought to understand the methylation patterns of CpG islands in
normal patients compared to AML patients. CpG islands are regions of the gene with unusually
large numbers of CG nucleotides. Cancer cells often contain hypermethylation of tumor
suppressor genes, which enables their proliferation; the suggested reason is due to elevated levels
of methyltransferase in leukemia (Melki et al. 3730). Certain subsets of leukemia contain
hypomethylated sites, and others contain hypermethylated cites. However, a large scale CpG
island methylation study across several genes in one subset of leukemia has not been studied;
furthermore, the mechanism for unusual methylation is not understood. Melki was particularly
interested in assessing whether this phenomena was a general process or limited to particular
genomes.
To study aberrant methylation, the team extracted bone marrow samples from twenty
patients with AML and 9 patients with no evidence of leukemia. The group could have been
better controlled, with age ranges varying widely; however, the authors’ data suggests that
methylation patterns of normal vs AML patients was conserved regardless of the age of the
patient. Bisulfite genomic sequencing, which involved the PCR amplification of many
implicated genes and was used to construct methylation maps, enabled comparisons in
methylation patterns for each patient across several genes.
The genes the authors chose were calcitonin, ER, E-Cadherin, p15, p16, HIC1, Rb, and
GST-Pi. Their reasoning for picking these genes was well justified. The first three have been
shown to be hypermethylated in many forms of acute leukemia through Southern blot analysis.
p16, a similar analogue of the cyclin dependent kinase inhibitor p15, did not have the same
property. E-cadherin, Rb, and GST-Pi has been shown to be hypermethylated in other cancer cell
lines. HIC1 contains an unusually high number of CpG islands even in intron regions, thus
enabling them to see how this property differs between introns and exons. The set of genes the
authors select contain a good number of positive controls and interesting variations that add
worth to their paper; it would however have been beneficial to isolate a negative control, such as
a hypomethylated gene. Furthermore, the functional roles of these proteins are seldom explained,
save for p15 and p16. Why might a cancer cell hypermethylate and thus silence these particular
genes? In particular, the calcitonin gene and cadherin gene are shown to be calcium dependent,
and as will be shown by their data, both genes are hypermethylated in AML patients. These
relationships may have been an important stepping stone forward to characterize AML further,
but are not discussed by the paper.
The gene maps (Figures 1-6) for each gene are similarly organized. The label A
highlights the CpG sites on the gene by lines and shows which range of CPG sites were studied
in the genome. B marks the sequence of the CpG island. C shows the methylation map of each
CpG site in N and R patients; N patients are normal, and R patients have AML. The shading of
the boxes in each map correspond to the percent methylation; dots in the boxes indicate that the
site was not determined. A representative version of these figures is shown in Picture A, which is
the gene map for p15.
In the case of calcitonin, ER, E-cadherin, p15, and HIC1 (Figures 1-4, 6), much more
methylation seems to be present in a select number of AML patients compared to the normal
patients. Nonetheless, each figure illustrates the heterogeneity of methylation patterns from one
patient to the other. In Picture A, the red arrow points to the middle patients with AML, who
have almost no methylation in their CpG sites. Right above and below the arrow are AML
patients who have extensive methylation in nearly every CpG site. The authors do a good job
presenting caution for their results, pointing out this feature instead of overemphasizing the
higher general methylation in AML patients. Nonetheless, the methodology has some problems.
First of all, the authors overemphasize that CpG sites near the promoter are more likely to
be methylated. This is seen clearly in the calcitonin region (Figure 1), but as seen in Picture A,
there is no seeming pattern between proximity to the promoter (marked by the arrow) and the
methylation of the CpG site. Not every site has a clearly defined transcription site, making this a
difficult point to assess across various genes. Another issue is that the region chosen for the CpG
site analysis seems rather arbitrary. In Picture A, the authors select a large portion of the gene
containing the intron and exon portions of the gene. However, in the calcitonin gene and the p16
gene, a much smaller portion is selected. It is not made clear why these discrepancies exist; as
CpG sites are ubiquitous in the genome. It would have been more prudent to study every CpG
site in every CpG island, as the heterogeneity of the data is a defining feature of the result.
The authors sum up the methylation results for each gene in Figure 7 (Picture B). For
each patient, they shade in the circle corresponding to a gene if at least 25% of the CpG sites are
methylated or if there is at least one site with greater than 25% methylation. This is a rather
arbitrary distinction to draw the line from, though it does ensure a cross-understanding of patient
variances for each gene. As is clear, most AML patients have at least one gene that is
hypermethylated, and the proportion of patients who meet this criteria are significantly more than
the normal group for most genes. The results also add context to other properties of genes; some
genes which have been shown to be hypermethylated in other cancers, such as Rb and GST-Pi,
are not hypermethylated in leukemia, and there is a difference in the hypermethylation of the
introns of HIC1 compared to the exons in the normal and AML group. Ultimately, these results
do confirm that hypermethylation of the CpG islands in several genes is a defining feature of
AML in contrast to normal patients. There is substantial heterogeneity between various patients,
suggesting that CpG methylation is non-specific.
Aesthetically, the paper is a bit cluttered and difficult to read. For instance, Picture A
presents the data for every one of the 47 CpG islands. It would have been worthwhile to
incorporate the data in a different way, such as percentage of sites showing a certain ratio of
methylation. Because figures like A are repeated six times, the paper feels longer than it needs to
be and is difficult to contextualize with such a massive number of data points. Picture B does a
good job contextualizing each gene for each patient, though not every single dot is needed to
make the point.
PICTURE A. Figure 4 from Melki et al. 1999. Figure 1 legend is included since it contains the
legend for understanding the shading patterns of the methylation map (C). Red arrow highlights
patients with AML who have similar methylation patterns as normal patients.
PICTURE B. Figure 7 from Melki et al. 1999.
III. Loss of acetylation at Lys16 and trimethylation at Lys20 of histone H4 is a common
hallmark of human cancer (Fraga et al. 2005)
Fraga et al. (2005) contextualize epigenetic studies in a variety of cancers and focus
particularly on histone modification, a less characterized (at that time) epigenetic feature of a
variety of cancers. Referencing the heavy focus on aberrant DNA methylation of CpG islands in
past research, the authors provide an interesting transition from the previous paper. In particular,
the authors consider methylation and acetylation at histone 4 (H4) in a variety of cells. Because a
substantial number of their experiments study leukemia and lymphoma, and because their
ultimate conclusion is that certain histone modifications are cancer “signatures” (ubiquitous
along most cancer cells), these findings could likely be extended to AML, with some
reservations to be discussed later.
The authors use a variety of methods to study the presence of histone modifications.
High-performance capillary electrophoresis (HPCE) produces spectra corresponding to a variety
of histone modifications; the area under the curve represents the frequency of these sites. Picture
C corresponds to Figure 1 of their paper and shows the differences between the normal cell line
of lymphocytes (NL) and a leukemia cell line (HL60). As is shown in the boxed area, the ratio of
tri-methylated H4 and single acetylated H4 falls in HL60 compared to NL. The authors confirm
these findings in a variety of cancers, and show that more advanced stages of cancer have more
advanced loss of these modifications (Figure 2/Picture D). Figure 2 as a whole is excellent,
accounting for a variety of factors, such as freshly extracted tumor cells, cell cycle controls, and
a variety of cell lines. The only issue is that in Figure 2A, it is difficult to to assess the statistical
reliability of the acetylated data. With these findings, the authors transition into the next logical
step- what amino acid is affected within the histone for each modification?
The problems of the paper become apparent in the next figures (3 and 4). The authors use
western blot analysis to pull down specific acetylated sites of H4, such as K5, K8, K12, and K16,
as well as the trimethylated K20. Excerpts from each figure is shown in Picture E. The authors
justifiably show that the acetylation of K16 decreases in the lymphomas, but do not explain the
apparent hyperacetylation with the other lysine residues. In figure 4, the authors show that there
is a decrease in TrimeK20H4 for two of the four lymphoma cell lines, but do not mention why
the Jurkat cells or Column 1 are apparently similar in their amounts of trimethylation. The
bottom part of 4A shows that the HPCE quantification results in a decreased amount of 3Me, but
the staining of the Jurkat lymphoma is inconsistent, with the right cell expressing far less
3MeK20H4, but the authors do not explain this finding. It is likely that the left cell is a control
and the right cell is a distorted cancer cell, but this is not indicated at all in the paper. By not
addressing these significant findings, the authors lose some legitimacy in trying to overstate the
universality of these phenomena across a variety of cancer cells.
After isolating K16 and K20 as the important sites, the authors transition into identifying
links between genes and these modifications. They begin first with CDKN1A, a gene which is
epigenetically silenced by hypoacetylation at the promoter. Comparing the histone modifications
of NL and HL60 cells (through a CHIP assay) shows no apparent change in trimethylation or
acetylation at lysine 16, though other histone modifications differ between the two cells. The
reasoning for choosing to study CDKN1A (or any histone-modified gene) is not explained well;
in the figure legend, the authors suggest that they are expecting a shift in the acetylation pattern
to K16, but it seems that this shift has already been identified previously as the other acetylated
histones as the difference between cancer and normal cells. It would have been better to study a
gene which is epigenetically regulated in the same way through histone modification in both the
normal and cancer cells.
The authors continue linkage studies by studying genes which have hypermethylated
CpG sites in HL60 cells but not in normal lymphocytes: MGMT and PRDM2. They use a
negative control GSTP1 and a positive control IVD to ensure that their methylation studies are
consistent, which is good experimental design, and find that both HL60 and NL cells have the
same K16 acetylation patterns and trimethyl K20 (Picture F). However, as previously indicated,
they also show acetylation patterns across the two cell lines, and is seen, they are clearly
underexpressed in the HL60 cells (Picture F). These results are not explained at all nor are they
made significant, even though they are anything but insignificant. If the authors propose that
there is no association between the studied histone modification and the gene, it would have
made a stronger argument if they did not have another histone modification which does appear to
be important. Similar to design as the above two experiments, the authors do show an association
of hypomethylated repetitive sequences with their histone modifications of choice, but they
continue to incorporate other histone modifications which are also significant, yet not studied in
the paper. If the goal of the paper is to isolate signature histone modifications of cancer, it would
have been important to study the ubiquity of these trends across a variety of cell lines.
The above issue touches with a significant consideration of the paper. The authors intend
to highlight a common phenomena across a variety of cancers, and they do so successfully,
clearly showing decreased levels of K16 acetylation and K20 trimethylation of H4. However,
they focus almost exclusively on leukemia and lymphoma, which are derived from a linked
immune-lymphatic system. It would have been more fair to suggest that these features are a
hallmark of lymph derived cancers, rather than as a “cancer signature” or the general “human
cancer”. Furthermore, the inclusion of other significant histone modifications that clearly show
up in a variety of genes might be more characteristic of a cancer signature across a variety of
genes.
Aesthetically, the paper is in general well organized and logically structured. The figures
are generally strong and make a strong case for the author’s main points, though comparisons
could benefit from a t-test. The major issue with the paper, however, is that there are a variety of
confounding factors in the data that are not expanded upon in either the results or the conclusion.
A minor issue is that not all of the experiments seem as important to make the point that the
authors intend. For instance, not outlined above is a variety of demethylating drug studies done
at the end of the paper to show that there is an increase in hypomethylation (Figure 8) in addition
to the already existing hypomethylation. This data does not seem to be very important in
establishing the importance of the histone modifications.
PICTURE C. Figure 1 from Fraga et al.
PICTURE D. Figure 2 from Fraga et al.
PICTURE E. Figure 3 (left) and part of Figure 4 (right) from Fraga et al. In 3A, the authors
correctly show that the acetylation of K16 is decreased in lymphomas compared to the normal
lymph cells, but they make no mention of the hyperacetylation of K12 or K5 (red circles). 3B is
not explained in the paper at all beyond the figure legend. In 4A, the trimethylation decreases in
lymphoma cell lines 2 and 3, but not in 1 or the Jurkat cell lines (red circle). The authors do not
mention this discrepancy. Furthermore, in their immunofluorescence of the K20 site, two Jurkat
cells vary widely in their relative amounts of H4-Lys20.
PICTURE F. Excerpt of Figure 6 from Fraga et al. The red boxes indicate acetylation levels of
K5, K8, and K12 in NL and HL60 cells.
IV. microRNA-29a induces aberrant self-renewal capacity in hematopoietic progenitors,
biased myeloid development, and acute myeloid leukemia (Han et al. 2010)
The next paper, by Han et al. (2010), is the most recent of the three papers studied here
which focus on epigenetic trends and patterns of acute myeloid leukemia: particularly of
microRNA (miRNA). As the authors mention, miRNAs have been linked to carcinogenesis by
affecting a variety of important normal processes such as apoptosis and proliferation. The
author’s unique methodology from other studies of miRNA is that they are studying the role of 1
particular miRNA and its role in directing differentiation of stem cells. Unlike Melki’s paper,
which looks particularly at AML, the authors study an extensive number of cell types, including
stem cells, cancer stem cells, immature progenitors, and mature progenitor cells. While this is
arguably the most difficult to read paper of the five due to its extensive use of a variety of cells
and complex flow-cytometry data, it presents a very strong case that microRNA-29a is
overexpressed hematopoietic stem cells and leads to a biased differentiation pattern which
ultimately results in acute myeloid leukemia. This paper is important because it highlights a
potential mechanism through which AML can develop. Unlike treatment options which look for
ways to reduce AML potency, treating for the microRNA and inhibiting it before it manifests
may be a viable option to prevent the emergence of AML.
To break down the paper in easier to understand terms, I will use the terms HSC to refer
to hematopoietic stem cells, LSC to refer to leukemia stem cells, MPP to refer to multipotent
progenitors (which are more mature than stem cells but not as mature as the next group), and
CLP to refer to lineage committed progenitor cells. The authors describe how they isolated
miRNA29a from a QT-PCR study of 315 RNAs. They very clearly show that miR-29a is more
expressed in HSCs, MPPs, and LSCs than in CLPs through heat map expression (Picture G).
However, in relative expressions in human and mice cells of miR-29a, the bottom half of Picture
G, their data is quite confounding. While both cells show an overexpression of 29a in HSCs and
an under expression of 29a in CLPs, their MPP data is different; one is significantly lower, while
the other is not. Furthermore, LSC and non-LSC cells are not studied in the mouse genome. It is
not clear why the authors chose to do their methodology that way.
Next, the authors transduce normal BM mouse cells with either 29a or a control EMP.
Through flow cytometry, the authors show that 29a transduced cells and spleen cells show
myeloid lineage bias by overexpressing monocytes and granulocytes (Picture H). The data is
confusingly organized. It seems that the left four boxes are the EMP control, and the right four
boxes are the 29a. But the third column corresponding to the 29a is the weakest expressing of
them all. The authors also use staining to show that there is a myeloid lineage bias in the wild
type cells compared to the chimeric miR-29a cells, though an arrow pointing to different cells
would have made for a better figure. However, contrary to what is expected, there is an increase
of CLPs as well. It makes sense that the levels of HSCs or MPPs will rise up in the cell after 29a
transduction, but the data before doesn’t seem to support that there would be an increase in
CLPs. Thankfully, this confusion is cleared up in the next figure, but it weakens their argument
and should have been addressed when it was seen.
In their next set of experiments (Picture I), the authors show that there is a rise in the
numbers of MPP in 29a cells, while the numbers of common myelin progenitors and granulocyte
myelin progenitors remains the same. The proliferative ability of MPPs is thus enhanced by 29a
transduction and can be the reason for a biased lineage. Logically, their next experiment is to see
what the levels of differentiated cells look like, and they see a rise in macrophages and
granulocytes, as well as a corresponding decrease in megakaryocytes. AML is a myeloid
disorder, and myeloblasts are important in producing granulocytes. Furthermore, leukocytes
include macrophages and granulocytes; erythrocytes and thrombocytes are not part of the
leukocytic pathway. While this is not clearly mentioned in the paper, the lineage bias points
towards elevated levels of myeloid progenitors, which is characteristic of myeloproliferative
disorder (MPD). There is also a decreased expression of T-cells and B cells (Picture I, part D),
which are leukocytes as well, but are not seen in leukemia, suggesting that a particular branch of
leukocytes are upregulated. As the authors logically address in their next figure, the question
becomes why these particular progenitors are favored, and of what implications they have for
leukemic stem cells.
After a few months of transplantation with the overexpressed granulocyte and
macrophage progenitors, the cells show splenomegaly and hepatomegaly; they also are more
densely packed with immature myeloid blasts, a hallmark feature of AML (Picture J, part A).
Furthermore, AML cells have higher levels of mi-R29a. This suggests that the myeloproliferative
disease has evolved into AML, and that the progenitors have become capable of self-renewal,
which is a concerning implication of how AML might initiate. Mature progenitors are not
normally capable to self-renew, so it may be that AML adds this element to them and enables
them to revert back to immature blasts which can continue to create these self-renewing
progenitors. This is the most substantial and striking data point that they have, and it adds a new
lens to consider in AML research of the importance of the progenitors themselves. As
highlighted in the introduction, it has been demonstrated that AML is the result of immature
myeloblasts which interfere with normal cells. It may be useful to study whether the self-