Cancer research with non‐coding RNA

Abstract

Cancer research is not limited to medical research; it expands over several disciplines, incorporating molecular bioscience at both the macro and micro levels. All stages and aspects of cells, from development and differentiation, apoptosis, cell adhesion and many more, are research fields with a connection to cancer. Cancer research in itself is the research of cancer cures. Recently, not only cancer but also bioscience research has surfed on the new wave of RNA knowledge. Most of those RNAs are non‐protein‐coding RNAs and are connected to cell development and differentiation, and thereby with cancer differentiation and treatment. Here we would like to introduce the latest in cancer research that has emerged from the field of molecular biology research. (Cancer Sci 2006; 97: 1285–1290)

Functional non‐coding RNA

Until recently, it was commonly thought that the mRNA transcripts transcribed from the genome were translated to protein, with the exception of tRNA, rRNA and snoRNA, which function as RNA molecules. Basically, mRNA was the mediator step in protein biosynthesis from genome information to a finished protein. However, in 2002, the Functional Annotation of Mouse (FANTOM) consortium found several mRNA transcripts that were not translated and seemed to function as RNA molecules.^{( 1 )} This was the first clue that the old concept of Central Dogma (DNA transcript to RNA, RNA translate to Protein) did not measure up to the complexity of the reality. In 2005, the FANTOM consortium found even more evidence for these non‐protein‐coding RNA transcripts when ∼103 000 cDNA sequences were classified as non‐coding RNA (ncRNA).^{( 2} , ^{3 )} This large amount may suffer from sequencing and cloning errors, but even when accounting for errors, this number is too large to ignore, and further analysis is needed. However, it is clear that the organisms are more complicated than we believed and that functions of RNA are still concealed.

Before discovering the RNA‐mediated post‐transcriptional gene silencing (PTGS) that uses non‐coding RNA, the main protein for gene regulation consisted of transcription regulation factors (TRF). In this regulation system, TRF bind target promoter sequences on genes, thereby regulating their expression by recruiting the molecules necessary for transcription. Negative expression (suppression of transcription) is achieved by altering the promoter or enhancer sequences by methylation or chromatin (histone) modification or by targeting the TRF itself, everything in order to stop the TRF binding the promoter or enhancer sequence.^{( 4} , ^{5 )} With the recent findings connected to PTGS, some of these chromatin and histone methylation changes have proven to be under the control of PTGS, revealing a tight interaction network between both systems of gene regulation. The transcriptome has turned out to be a complex network with hidden players. With the discovery of the RNA continent, one of the players has stepped into the light. These findings will affect not only molecular science; the repercussions will spread all over the biological sciences and medicine in particular. Below we will introduce the newly found details of functional non‐coding RNA, the post‐transcriptional regulation by RNA and the relationship between the function of micro RNA (miRNA) and the onset of cancer.

Small interfering RNA

The small interfering RNA (siRNA) transcripts are ∼21–25 bp long, double‐stranded transcripts that were found in Caenorhabditis elegans in 1998 as a defense response to non‐endogenous double‐stranded RNA (dsRNA), leading to sequence‐specific mRNA cleavage.^{( 6} , ^{7 )} This double‐stranded induced cleavage was named RNA interference (RNAi) and was soon to be found in other organisms, for example Drosophila melanogaster,^{( 8 )} and in vertebrates.^{( 9 )} The main function of the mechanism is believed to a defense system against introduced viral double‐stranded sequences, cutting them to pieces and thereby rendering them unable to infect the cell. The gene regulation part was not discovered until recently. RNAi was also quickly adapted for use in the laboratory as a gene‐silencing tool for the silencing of specific genes by introducing the specific double‐stranded sequence, and has today become a common tool for the study of gene expression and regulation due to its specificity^{( 10} , ^{11 )} (Fig. 1). siRNA is generated by Dicer cleavage of long double‐stranded RNA sequences.^{( 12} , ^{13 )} After cleavage, the double‐stranded siRNA is loaded into the RNA‐induced silencing complex (RISC).^{( 14 )} During the loading process one strand is peeled off, leaving a single‐stranded siRNA lodged inside RISC. This strand is then used by RISC as a template to recognize the cleavage target in the target mRNA transcript. A full match induces a cleavage − in Drosophila RISC complexes, Argonaute2 has been identified as carrying the slicing action, but several other proteins are also involved in the RISC complex. A less than full match induces translational depression.^{( 15} , ^{16 )}

Figure 1.

There are various kinds of small interfering RNA (siRNA) that cause gene expression suppression. Most commonly used is synthetic siRNA, and the siRNA expression vector. The bidirectional 2‐promoter expression vector has two RNA polymerase III promoters located on both sides of the siRNA sequence, which is complementary to the sense and antisense transcript. The short hairpin RNA (shRNA) expression vector has one RNA polymerase III promoter at one side of a sequential arrayed sense and antisense in order to form a hairpin structure. Transfected or expressed siRNA is captured by the RNA‐induced silencing complex (RISC) where one strand is removed. The guide RNA transcript and RISC complex are then ready to recognize and cleave a target mRNA. dsRNA, double‐stranded RNA.

Complete siRNA biogenesis is still not fully described; the latest news in RNAi is the sense/antisense (S/AS) endogenous origin of siRNA. This reveals even more possibilities for gene regulation, with an extra level of regulation in the creation of long double‐stranded RNA transcripts by S/AS transcription. Known S/AS pairs have been identified through the mapping analysis carried out in mice by the FANTOM consortium, to a number reaching 4520 for exon–exon cis S/AS mRNA pairs and 4129 for exon‐overlapping cis S/AS pairs, numbers that most probably will increase in the future with the addition of trans S/AS pairs.^{( 17 )}

In addition to siRNA, which is a non‐coding RNA, other RNA transcripts have been revealed to be involved in RNAi or to have a similar function, for example trans‐acting siRNA (tasiRNA), repeat‐associated small interfering RNA (rasiRNA) and small scan RNA (scnRNA) (see Box and Fig. 2).

Table Box .

Trans‐acting small interfering RNA

Trans‐acting small interfering RNA has so far been found only in plants and nematodes. Compared to siRNAs that cause post‐transcriptional gene silencing of transposons, viruses and transgenes, the targets of tasiRNA are endogenous transcripts. Their biogenesis involves molecules from both the miRNA and the siRNA biogenesis pathways. TasiRNAs are genetically defined at specific loci and arise by phased, DICER‐like processing of dsRNA formed by RDR6/SGS3 activity on RNA polymerase II transcripts. TasiRNAs interact with target mRNAs and guide cleavage by the same mechanism as do plant miRNAs. The tasiRNA endogenous function is uncertain, as the function of its target genes remain unknown.( 4 )

Repeat‐associated small interfering RNA

Repeat‐associated small interfering RNA were first found in plants, flagellates (Trypanosoma brucei), fly (Drosophila melanogaster) and fission yeast (Schizosaccharomyces pombe). In yeast, rasiRNA derive from the Dcr (yeast Dicer) cleavage of long dsRNA derived from repetitive sequences transcribed in both sense and antisense directions. RasiRNA is the guide transcript in the RNA‐induced initiation of the transcriptional silencing complex (RITS), which consists of Ago1, Chp1 (a chromatin binding and histone methylase) and Tas3 (a histone methylase). This complex never leaves the nucleus and binds to DNA where it induces histone and DNA methylation, a process that is also known as transcriptional gene silencing (TGS), and is involved in the establishment of heterochromatin.( 35 , 36 , 37 ) In plants, the biogenesis of rasiRNA requires the RNA‐dependent RNA polymerases RDR2 and SDE4, the plant DCL3 (Dicer‐like) and the plant AGO4. There are some small differences in the final effects so far described for rasiRNA gene silencing in yeast compared to plants. Besides the establishment of heterochromatin, rasiRNA plus RITS is also able to repress repetitive transposable elements. In plants, rasiRNA is able to induce systemic silencing, to mediate histone H3 methylation and use asymmetric DNA methylation to repress mobile genetic elements.( 38 , 39 , 40 ) In D. melanogaster, it is suggested that the RNAi silencing components of the Argonaute family (piwi and aubergine) and spindle E (a RNA helicase) are involved in rasiRNA biogenesis, as mutations in these genes cause defective histone H3 Lys9 methylation and HP1 (a chromatin binding protein) associations in heterochromatin regions.( 41 )

small scan RNA

Small scan RNA is another interesting ncRNA species. It is found in Tetrahymena thermophila, a unicellular organism. The scnRNA is responsible for histone methylation leading to DNA elimination during cell conjugation (the reproductive phase of T. thermophila).( 42 ) Basically, the process serves to get rid of redundant DNA after a conjugation. T. thermophila possesses two nuclei, a somatic (household genes), diploid, macro nucleus and a germinal, haploid, micronucleus. During conjugation the two haploid micronuclei fuse in order to form a zygotic nucleus. The zygotic nucleus will later divide into two nuclei, which will differentiate into new micro and macronuclei. The differentiation requires the deletion of redundant DNA; the scnRNA not able to pair with the complementary internal eliminated segments (IES) in the old macronucleus diffuses to the new macronucleus. In the nucleus the scnRNA binds complementary IES sequences and cause them to be methylated and thereby singled out for elimination. This process is similar to siRNA and miRNA in that it requires the presence of an Argonaute family member (Twi1) to eliminate the transcripts and the presence of Dicer in order to produce the scnRNA. The elimination is not a small feat, as the eliminated sequences amount to ∼6000 fragments of different lengths, all in all ∼15% of the total genome is removed.( 43 , 44 , 45 )

Figure 2.

Non‐coding RNA biogenesis. Small interfering RNA (siRNA), trans‐acting siRNA (tasiRNA) and repeat‐associated small interfering RNA (rasiRNA) all derive from long double‐stranded RNA transcripts. siRNA comes from endogenous or exogenous (e.g. viral) sources through processing by Dicer into ∼21–22mer long transcripts. tasiRNA uses parts from the biogenesis of both siRNA and micro RNA (miRNA), and direct cleavage of endogenous cognate mRNAs in trans and has only been found in plants and nematodes. rasiRNA originates from long repeat double‐stranded RNA transcripts and therefore target repeats, and is involved in the establishment of heterochromatin in repetitive elements. The rasiRNA effector complex is the RNA‐induced initiation of transcriptional silencing complex (RITS). Small scan RNA (scnRNA) is likely to use some parts of miRNA and siRNA biogenesis as Twi, an Argounaute protein, is required together with the scnRNA for H3K9 methylation and subsequent elimination of DNA. RISC, RNA‐induced silencing complex.

siRNA and cancer.  There are still unknown details regarding the biogenesis and function of siRNA, especially in mammals. The siRNA technique of using siRNA transcripts to knock down gene expression of specific genes has quickly become a popular method used for gene function analysis in molecular biology. The method is still being developed to improve the siRNA vectors, both the specificity and the usability of the technique, but already exhibits a high specificity. The hope of the future is to be able to use siRNA knock down techniques to treat both genetic disorders, viral infections and, in extension, to be able to use siRNA as a cancer treatment (see Table 1).^{( 18} , ¹⁹ , ^{20 )} Epigenetic modifications of DNA and histones are also influenced by rasiRNA, and are relative to gene expression. Abnormal gene expression causes several clinical disorders, and abnormal epigenetic mechanisms are connected to developmental failure and several cancer forms. The scnRNA that causes a dynamic genome rearrangement and DNA elimination is very specific to Tetrahymena, and has not yet been found in any other organism. If a similar scnRNA system were to be detected in mammals, it may be possible that it will be related to loss of heterozygosity or chromosome fusion.

Table 1.

Cancer‐associated genes targeted by RNA interfernce

Pathway	Target genes
Apoptosis	BAX
	BCL2
	p53
	HDM2
	p21
Signal transduction	H‐Ras
	K‐Ras
	PLK1
	TGF‐β
	STAT3
	EGFR
	PKC‐a
	FAK
Viral oncogenes	EBV LMP‐1
	HPV E6, E7
Fusion genes	BCR–ABL
	EWS–FLI1
Others	Telomerase
	Fatty acid synthase
	CytosinB
	β‐Catenin
	VEGF
	FGF4
	MDR

miRNA

In contrast to siRNA, miRNA are mostly known to be involved in biogenesis and cell function mechanisms, and several miRNAs are related to development^{( 21} , ²² , ^{23 )} and differentiation (see Table 2). For example, embryonic development in the zebrafish requires miR‐140 (head skeleton), miR‐124a (nervous system), miR‐206 (muscles) and miR‐126 (blood vessels).^{( 24 )} More than 450 miRNA have now been reported in humans (http://microrna.sanger.ac.uk/sequences/), and more than 1000 miRNA have been predicted.^{( 25 )}

Table 2.

Development‐ and differentiation‐related micro RNA (miRNA)⁽ 29, 30 ⁾

miRNA	Function	Known targets	Species
lin‐4	Developmental timing	lin‐14, lin28	Ce
let‐7	Developmental timing	lin‐41, lin‐57, Ras	Ce, Hs
lsy‐6	Neuronal patterning	cog‐1	Ce
miR‐273	Neuronal patterning	die1	Ce
Batam	Cell death, proliferation	hid	Dm
miR‐1	Heat differentiation, myogenic	hand2, HADAC4	Mm
miR‐10a	Megakaryocytopoiesis	HOXA1	Hs
miR‐14	Cell death, fat storage	caspase, Bcl2	Dm, Hs
miR‐15	Cell death	Bcl2	Hs
miR‐122	Regulation of lipid metabolism	?	Mm
miR‐130a	Megakaryocytopoiesis	MAFB	Hs
miR‐133	Skeletal muscle proliferation	SRF	Xl, Mm
miR‐134	Synaptic development	Limk1	Rn
miR‐143	Adipocyte differentiation	?	Hs
miR‐181	Haematopoiesis	?	Mm
miR‐196	Development	HoxB8, HoxC8, HoxD8, HoxA7	Mm
miR‐375	Insulin secretion	Myotrophin	Mm

Ce, Caenorhabditis elegans; Dm, Drosophila melanogaster; Hs, Homo sapiens; Mm, Mus musculus; Rn, Rattus norvegicus; Xl, Xenopus laevis.

Micro RNA is very similar to siRNA, and both transcripts are used as guide transcripts in the effecter complex RISC. The main difference lies in the biogenesis; whereas siRNA originates from long double‐stranded RNA, miRNA originates from the double‐stranded part of RNA hairpins. miRNA also requires additional processing and passes through several maturation steps; the first cut is done by Drosha, which cuts away the hairpin ‘feet’, followed by export to the nucleus where the loop is cut off by Dicer.^{( 26} , ^{27 )} Both siRNA and miRNA are double stranded before incorporation into RISC, but upon insertion one strand is removed.^{( 28} , ^{29 )} The target of miRNA‐induced RNA silencing is usually located in the 3′ untranslatable regions (UTR) and one miRNA can target more than one gene's UTR, giving miRNA a somewhat lower specificity but wider range of targets than siRNA. Several miRNAs can also have the same target, giving rise to a complex regulation network of miRNA and miRNA targets. When the miRNA mismatches with the target sequence, it will cause translational depression. When it is fully matched, it will cause cleavage of the target mRNA. Both actions will look the same; a lowering of expression rate, but in the first case the mRNA translation is delayed (sometimes until it degrades), whereas in the second case it is made impossible by destruction of the transcript (see Fig. 3).

Figure 3.

Micro RNA (miRNA) originates from long primary hairpin transcripts transcribed from genomic DNA by RNA polymerase II. These primary transcripts (pri‐miRNA) are cleaved by Drosha (limited to animals, plants use DCL1 instead) plus cofactors to yield pre‐miRNA with a ∼50–70‐nucleotide‐long hairpin loop structure. This pre‐miRNA is exported by Exportin‐5 to the cytoplasm and cleaved by Dicer to yield the 18–24‐nucleotide‐long miRNA. When the miRNA is incorporated into the effector complex, RNA‐induced silencing complex (RISC), one strand is degraded and the remaining strand is used as guide transcript. A partial match to the target induces translational depression whereas a complete match induces cleavage of the target mRNA. Plant miRNA tend to have a higher grade of complimentary to the target sequence and therefore induce cleavage to a higher degree.

miRNA and cancer.  Because miRNA is involved in cell differentiation and development, it is an obvious focus for any study of cancer and cancer development. Similar to coding genes, miRNA also possesses tissue specificity, differentiation, developmental timing and specific expression. Recent research has indicated intimate links between miRNA expression and cancer differentiation, and it is clear that loss of heterozygosity (which is connected to tissue specificity in cancer) in loci with genomic deletions or duplications can disrupt several miRNA coding regions. These results indicate that genomic deletion or duplication influences not only the gene expression level in those loci, but also targets the expression level of miRNA located in those genomic regions. This gives new possibilities in diagnosing tissue‐specific cancers with the help of miRNA transcripts and their location in loci affected by loss of heterozygosity. Additionally several miRNAs have already been connected to different cancer types and cancer stages (see Table 3).^{( 30} , ³¹ , ^{32 )} In functional analyses, miRNA is easier to clone and determine the sequence of compared with siRNA, and large databases have already been collected. This large amount of data makes it easier for scientists to predict miRNA and its probable targets and possible cancer connections. Microarray analyses have further increased the amount of data available on miRNA from cancer tissues and have simplified the search for connections between cancer oncogenesis and miRNA expression.^{( 33 )}

Table 3.

Cancer‐related micro RNA (miRNA) transcripts

Cancer	Expression level in cancer	miRNA ID	Mapping position	Known target gene of miRNA
Blast cancer	Up	miR‐21	17q23.2
		miR‐155	21q21	AGTR1
	Down	miR‐10b	2q31
		miR‐125b1,b2	11q24.1, 1q11.2	lin‐28
		miR‐145	5q32.33
Glioblastome	Up	miR‐10b	2q31
		miR‐130b	11q12
		miR‐221	Xp11.3	KIT
		miR‐125b1,b2	11q24.1, 1q11.2	lin‐28
		miR‐9–2	5q14miR‐21	17q23.2
		miR‐25	7q22miR‐123	9q34
	Down	miR‐128	2q21
		miR‐181a,b,c	9q33.1‐q34.13, 1q31.2‐q32.1, 19q13.3	Hox‐A11
B cell chronic lymphocytic leukemias	Up	miR‐101	1p31.3
		miR‐10b	2q31
		miR‐123	9q34
		miR‐132	11q12
		miR‐134	14q32	Limk 1
		miR‐140	16q22.1	Histone deacetylase 4
		miR‐141	12p13
		miR‐153‐precusor	2q36
		miR‐154	14q32
		miR‐15b‐precusor	3q26.1	BCL2
		miR‐181b‐precusor	1q31.2‐q32.1	Hox‐A11
		miR‐183‐precusor	7q32
		miR‐188	Xp11.23‐q11.2
		miR‐190	15q21
		miR‐196–2	12q13	Hoxb8
		miR‐19a	13.q31
		miR‐217	2p16
		miR‐24‐1‐precusor	9q22.1
		miR‐33	22q13.2
		miR‐92–1	13q31
	Down	miR‐192	11q13
		miR‐213	1q31.3‐q32.1
		miR‐220	Xp25
Hepatocellular carcinoma	Up	miR‐224	Xq23
		miR‐18	13q31.3
	Down	miR‐199a1,a2	19p13.2, 1q24.3
		miR‐200a	1p36.33
		miR‐125a	19q13.41	lin‐28
		miR‐195	17p13.1
Papillary thyroid carcinoma	Up	miR‐146a, b	5q33.3, 10q24.32	traf6, IRAK1
		miR‐221	Xp11.3	KIT
		miR‐222	Xp11.3	KIT
		miR‐21	17q23.2
		miR‐220	Xp25
		miR‐181a,b,c	9q33.1‐q34.13, 1q31.2‐q32.1, 19q13.3	Hox‐A11
		miR‐155	21q21	AGTR1
		miR‐213
	Down	miR‐9–3	15q26.1
		miR‐219–1	6p21.32
		miR‐138–1,‐2	3p21.33, 16q13
		miR‐345	14q32.2
		miR‐26a1	3p22.3
Colon cancer	Up	miR‐223	Xq12
		miR‐21	17q23.2
		miR‐17	13q31.3
		miR‐106a, b	Xq26.2, 7q22.1
	Down	miR‐143	5q32	ERK5
		miR‐145	5q32.33
		miR‐195	17p13.1
		miR‐130a	11q12.1	MAFB
		miR‐331	12q22
Lung cancer	Up	miR‐21	17q23.2
		miR‐24–1(miR‐189)	9q22.32
		miR‐200b	1p36.33down
		miR‐126	9q34.3
		miR‐30a	6q13
		miR‐143	5q32	ERK5
		miR‐145	5q32.33
		miR‐188	Xp11.23‐q11.2
		miR‐331	12q22

Conclusion

Science moves forward, as always, but somewhat faster than usual, it seems, during the latest decade. This is especially true for the field of life science, where the advent of several technologies (e.g. polymerase chain reaction, 454‐sequencing and microarrays) have advanced our knowledge rapidly. The latest discovery is the enormity of the large RNA continent; the functionality of non‐coding RNA undermined our ‘common knowledge’. We have again had to realize that what we held as true was only true for a certain time, what we believe today will be history tomorrow. As researchers we can only try look forward and rebuild our knowledge as we go on. Our next step involves large possibilities for the development of new medical treatments and drug targets, with the accumulating knowledge of RNAi mechanism and ncRNA gene regulation. The biogenesis of the well‐known ncRNA species remains largely unknown, and we have only recently traced the connection between sense and antisense pairing to ncRNA and RNAi. We are charting unknown territories in the ‘RNA continent’, much in the same way that Columbus set sail for India − we do not know where we will end up.