Summary of ICMS Seminar Club presentation on Friday, March 25, 2022.
Title: Long non-coding RNAs (lncRNAs) in Cancer Development and Drug Resistance
Speaker: Prof. Eleonora Leucci, PhD, Laboratory for RNA Cancer Biology, KU (Katholieke Universiteit) Leuven, Belgium, and Head of TRACE (KU Leuven PDX platform)
On Friday, March 25, Prof. Eleonora Leucci gave a presentation at Fujita Health University. She talked about long non-coding (lnc) RNAs and how cancers can use them to their advantage.
Recording: For members of Fujita University, a recording of the meeting (without the discussion part) will be available at our Manabi system. Unfortunately, we cannot open the recording for a wider audience.
There were 41 participants, and the atmosphere was very good. Several people told me afterwards how much they had enjoyed the presentation and the fact that in our seminars a wide variety of topics is covered by speakers who are at the top in their field. Most of us (including me) had probably heard of lncRNAs before, but, before the presentation, had not realized the enormous potential of lncRNAs and their possible functions.
A nice slide that Prof. Leucci showed during her presentation, expressing her love for RNA, is shown in Fig. 1. After explaining the importance and versatility of RNA, and that various major discoveries on RNA had resulted in Nobel prizes, she concluded that “RNA rocks!“
The contents of the presentation
The below summarizes large (but not all) parts of Prof. Leucci’s presentation. Because the topic is quite complicated, I decided to summarize most of it in the form of questions and examples. I also added information from the literature.
What is RNA?
Ribonucleic acid (RNA) molecules essentially are strands of ribonucleic acid nucleotides. The core part is a negatively charged backbone of phosphorylated ribose sugars, each carrying one of the bases adenine (A), cytosine (C), guanine (G), or uracil (U).
Besides these core parts, RNA chains carry further modifications, as described in our seminar on January 25 this year by Dr. Wolfgang Leitner when discussing COVID-19 RNA vaccines.
RNA chains can be very short, such as several tens of nucleotides, but can also be many thousands of nucleotides.
The bases of the nucleotides have a tendency to form interacting pairs (A-U and C-G), so RNA stretches can bind to complementary RNA or DNA stretches. If those stretches are within the same RNA strand, this can lead to the formation of intramolecular nonlinear structures such as, for example, stem-loops.
What are the functions of RNA?
For discussing RNA functions in a general sense, it is important to understand that essentially it is just a material. Equally, it would be difficult to discuss “the function of proteins.”
As a material, features of RNA that are superior to those of proteins, are:
• Storage of biological memory. RNA strands can precisely copy the information on other RNA or DNA strands, and this information may be “translated” into proteins. Most of us are well familiar with coding messenger RNAs (mRNAs) and viral RNAs that encode proteins.
• Lengthiness. If comparing an RNA molecule and a protein of similar mass, the RNA molecule can be more stretched out, providing unique features as a scaffolding molecule.
• Selectivity of binding. The base-pair binding mode provides a versatile and (in evolution) more easily adaptable mode for selecting binding partners.
Many researchers believe that, at some point in evolution, “life” only existed in the form of RNA. A reason for such belief is that, even now, besides the above-mentioned properties, RNA can also execute functions that in present life-forms have mostly been taken over by proteins:
• Catalytic functions. For example, in ribosomes, which are the abundant small engines of the cell for mRNA translation and which for two-thirds of their mass exist of RNA (rRNA) besides also containing various proteins, the catalytic activity is largely performed by the rRNA part (Fig. 2A) (Youngman et al. 2004). Ribosomes organize the binding of transfer RNAs (small RNA molecules that carry an amino acid) to complementary nucleotide triplets on the mRNA, and promote the elongation of the encoded peptide chain by cleavage of the amino acid from the tRNA and the creation of a new peptide bond (Fig. 2B).
What kind of non-coding RNAs are there?
Humans have about 20,000 genes from which mRNAs are transcribed that encode proteins (Fig. 3).
Besides those protein-coding RNAs, there are a variety of non-coding RNAs, such as:
rRNA (ribosomal RNA; involved in mRNA-to-protein translation)
tRNA (transfer RNA; involved in mRNA-to-protein translation)
miRNA (micro RNA; involved in mRNA degradation)
piRNA (Piwi-interacting RNA; forming complexes with piwi-subfamily Argonaute proteins that are best known for silencing of transposable elements)
siRNA (small interfering RNA; involved in mRNA degradation; not naturally found in humans)
snoRNA (small nucleolar RNA; guiding chemical modifications of other RNAs)
snRNA (small nuclear RNA; involved in the processing of pre-messenger RNA)
lncRNA (long non-coding RNA; transcripts of >200 nt that do not encode proteins and do not belong to the other categories of non-coding RNAs)
What are long non-coding (lnc) RNAs?
LncRNAs are, by artificial definition, >200 nt and include a wide variety of transcripts. Some of them are transcribed from transcriptionally active coding gene regions, either as splicing variants, antisense transcripts, or intronic or intergenic transcripts. Others are from regions without coding genes. Most lncRNAs are expressed at very low levels, but some are expressed at high levels. LncRNAs typically are transcribed by RNA polymerase II, so that like mRNAs they are capped and poly-adenylated and undergo intron splicing. However, their transcription and sequences usually have not been optimized for these types of processing, and together with the fact that some lncRNAs have nuclear-retention signals, many remain in the nucleus (Guo et al. 2020). Some lncRNAs, however, are efficiently exported to the cytoplasm.
Prof. Leucci explained that some lncRNAs are also translated into small peptides, somewhat blurring the distinction between coding and non-coding RNAs (Choi et al. 2019). Her group is studying some of these lncRNAs, and also how lncRNAs can directly affect the translation machinery.
What do lncRNAs do?
LncRNAs can bind proteins, RNA, and DNA, and all kinds of specific effects have been reported for a large number of lncRNAs (Fig. 4; Zhang et al. 2019). Many lncRNAs remain in the nucleus, and a prominent mechanism through which lncRNAs exert their function is by interacting with transcription factors or chromatin-modifying complexes to modulate gene transcription (Quinodoz and Guttman 2014). Furthermore, RNA-driven phase separation is increasingly recognized as a major organizer of cellular compartmentalization by forming the scaffolds for “membraneless bodies.” The largest nuclear membraneless bodies that incorporate several specific lncRNAs are the nucleoli (e.g., they incorporate the lncRNAs PAPAS and SLERT; Pirogov et al. 2019). Other nuclear membraneless bodies that incorporate lncRNAs and use them as scaffolds for binding multiple proteins, include (Fig. 5):
• Barr bodies (which rely on the lncRNA XIST; see also below)
• Speckles (which incorporate the lncRNA MALAT1). Speckles seem to be involved in the organization of actively transcribed genomic regions. MALAT1 influences the incorporation of several speckle components but is not necessary for speckle formation (reviewed by Sas-Nowosielska et al. 2021).
• Paraspeckles (which rely on the lncRNA NEAT1-2). Paraspeckles seem not to be essential for cells, and several functions have been proposed (reviewed by Sas-Nowosielska et al. 2021).
• Gomafu bodies (which rely on the lncRNA GOMAFU). GOMAFU is believed to influence several neural processes. GOMAFU knockout mice show no developmental disorders but exhibit some unusual behaviors (reviewed by Sas-Nowosielska et al. 2021).
Also in the cytoplasm, mebraneless bodies are found that incorporate lncRNAs, including (Fig. 6):
• P-bodies (processing bodies). P-bodies are accumulations of RNA and proteins involved in mRNA degradation and miRNA silencing. P-bodies also include lncRNAs (Pitchiaya et al. 2019).
• Stress granules. Upon stress, cytoplasmic mRNA are sequestered into stress granules (SG) which are insoluble ribonucleoprotein granules. LncRNAs are part of these granules (reviewed by Campos-Melo et al. 2021).
Prof. Leucci emphasized that lncRNAs are important for cells to deal with stress, in part by the reorganization of cellular compartmentalization, and gave the example that already since the 19th century it is known that the number and size of nucleoli is increased in cancer cells (Derenzini et al. 2009).
How many lncRNAs are there?
In Fig. 3, you can see that Atianand and Fitzgerald, 2014, estimated that humans have about 14,000 lncRNAs. On average, the evolutionary life span of lncRNAs is shorter than that of coding genes, and most human lncRNAs are only shared with other primates.
Genome size, or even the number of protein-coding genes, does not increase with the complexity of the organisms. However, it has been published that the ratio of non-coding to protein-coding sequences in the genome increases with life complexity, with humans considered to be the most complex species (Mattick 2004). This has been interpreted as that lncRNAs (transcribed from the noncoding regions) may drive species complexity. However, especially in 2004, the quality of information may not have been good enough to make reliable and comprehensive comparisons between many species on this matter. Even if this assumption by Mattick 2004 would be true, it is not immediately clear why the noncoding/gDNA ratio, rather than the absolute size of the noncoding regions, would be relevant for the number and/or importance of lncRNAs.
It has been speculated that the number of lncRNAs is positively associated with the complexity of life, with us humans at the top end (e.g., Marín-Béjar and Huarte 2015). However, the depth and quality of transcriptome and genomic sequence information is very different per species, and it is questionable whether good comparisons can be made. Necsulea et al. 2014, who wrote a frequently referenced lncRNA quantification paper, wrote about comparison between species: “Although part of the variability in lncRNA repertoire size may be biologically meaningful, much is likely to be explained by unequal sequencing depth and by variable genome sequence and assembly quality.”
The audience at our seminar asked how species with a reduced genome size (e.g., Japanese pufferfish has a similar number of genes as human and zebrafish but its genome is 7.5 and 4 times smaller) are faring in relation to lncRNAs. Since many lncRNAs are from noncoding regions, would such species have a reduced number of lncRNAs, or would there be some form of compensation by having a higher density of lncRNAs in the remaining part of the genome? Prof. Leucci answered that, technically, lncRNA counting studies have too many open questions for being able to answer this question at this stage.
Are lncRNAs important?
Some lncRNAs are definitely important. The first lncRNA that was discovered was Xist (X-inactive specific transcript; Brown et al. 1991). The Xist gene maps to the X chromosome of placental mammals and in females acts as a major effector of, and is essential for, the inactivation of one X chromosome per cell. The Xist gene is only transcribed on one of the X-chromosomes, giving a 17 kb transcript which forms a complex with many proteins that spreads over the chromosome and induces epigenetic inactivations (Lu et al. 2017).
While most lncRNAs are poorly conserved in evolution, the lncRNAs MALAT1 and NEAT1 are well conserved throughout vertebrates. This suggests an important function, and, as described above, they have a pronounced effect on nuclear body formation. However, as Prof. Leucci explained to us, mice seem not affected when either of these genes is knocked out.
Interestingly, for cancer cells (as exemplified below), the importance of several lncRNAs is more readily found. As we discussed with Prof. Leucci, it may be that either only cancer cells use the respective lncRNA to their benefit, or that we still have to find the importance in healthy cells. I remind you that Prof. Hans Clevers also first identified some processes in cancer cells before he identified them in healthy (stem) cells.
There definitely exist unimportant lncRNAs that only are formed because of background transcriptional activity in normal cells or aberrant transcriptional activity in cancer cells. The discussion between researchers is about the proportion of functional versus nonfunctional lncRNAs, and/or whether the bulk of lncRNAs together may have some kind of buffering activity.
LncRNAs in cancer
The reason that Prof. Leucci started to study lncRNAs in cancer was that most mutations associated with cancer are in non-coding regions of the genome (Beroukhim et al. 2010; Cheetham et al. 2013; Tseng et al. 2014; Melton et al. 2015).
In the presentation by Prof. Hans Clevers on January 28 this year, we learned that many cancers only truly become malignant if besides a continuous switch-on of activation/proliferation pathways (e.g., by mutations in KRAS) the cell genome and transcriptome are “wildly shaken up” by mutations in the transcription factor p53 (leading to accumulation of DNA damage). The deregulation of normal control mechanisms in the cell also lead to an increased expression of many lncRNAs, and enhanced expression of lncRNAs can be used as a marker for the detection of cancer. Furthermore, these lncRNAs provide a reservoir of material that a cancer cell may use to enhance or deregulate processes. For example, it is important for a cancer cell to block apoptosis mechanisms that are induced by its cancerous state.
Fig. 7 is a figure by Aprile and Katopodi, 2020, that Prof. Leucci showed in her presentation. It summarizes (in red text) how many lncRNAs are found specifically associated with some cancers and have oncogenic properties. Reminiscent of how many lncRNAs are highly specific for some tissues (Mattick 2011), they also show high cancer-specificity (Brunner et al. 2012)
Whereas the importance of the lncRNAs NEAT1 and MALAT1 for mice remains to be discovered (see above), NEAT1 (Adriaens et al. 2016) and MALAT1 (Arun et al. 2016; Kim et al. 2018) have been described as very important for tumor initiation and progression. Prof. Leucci and her group found that NEAT1 expression is enhanced in tumor cells and they detected some mechanisms how this may help the cancer cells cope with stress, including the stress induced by therapy (Barra et al. 2020).
At least to some degree, lncRNA specificities for certain cancer cell types appear to be caused by cell-type specific chromatin condensation (inactivity) levels of different genomic regions in relation to their transcriptional activities. For example, the melanoma-specific lncRNA SAMMSON, that Prof. Leucci investigates (e.g., Leucci et al. 2016; Vendramin et al. 2018), is situated besides the gene for MITF (melanocyte inducing transcription factor); MITF is a transcription factor also expressed in healthy melanocytes (Fig. 8) and has been described as a “master regulator of melanocyte development and melanoma oncogene” (Levy et al. 2006). SAMMSON has not been detected in healthy melanocytes (Leucci et al. 2016). It is interesting that SAMMSON is only expressed in melanoma from the transition to becoming malignant (Fig. 9; Leucci et al. 2016).
The mechanisms of SAMMSON
Expression of the lncRNA SAMMSON is found in almost all melanoma. The mechanism of SAMMSON in supporting melanoma progression appears to be quite specific. Prof. Leucci found that SAMMSON associates in the cytoplasm with mitochondria (Leucci et al. 2016). Furthermore, she found by precipitation studies that this lncRNA forms complexes with the proteins P32, XRN2, and CARF (Leucci et al. 2016; Vendramin et al. 2018). P32 and XNR2 have similar functions but in different cellular compartments, as they promote ribosomal RNA (rRNA) maturation in the mitochondria and nucleus, respectively. Prof. Leucci and her group found that SAMMSON disrupts the binding of CARF to XRN2 and so promotes ribosome biogenesis in the nucleus; meanwhile, SAMMSON promotes the association of CARF with P32 which increases the delivery of P32 to the mitochondria and therefore also promotes the ribosome synthesis (and protein production) in the mitochondria (Fig. 10). The latter promotes oxidative phosphorylation, because this mitochondrial pathway depends on the translation, within the mitochondria, of the mitochondrial-genome-encoded OXPHOS proteins. Cancer cells, including melanoma, have an overall increase in anaerobic glycolysis and a decreased rate of oxidative phosphorylation; this is called the Warburg effect (Van der Heiden et al. 2009). It seems that SAMMSON puts a safety break to this by ensuring some level of oxidative phosphorylation. Prof. Leucci speculates that this is necessary to reduce apoptosis-inducing stress caused by mitochondria-targeting proteins that accumulate in the cytoplasm instead if there is not sufficient oxidative phosphorylation. An additional possibility is that sufficient usage of the oxidative phosphorylation pathway is necessary to produce some biomolecules needed by the cell. Either way, SAMMSON appears to reduce the stress of the cell and to protect it from apoptosis, which Prof. Leucci proved by showing that knockdown of SAMMSON in human melanoma cells in a mouse xenograft model increased apoptotic activity (measured by caspase3 activity) and reduced tumor growth (Fig. 11). Excitingly, in such models, she found that a combination of SAMMSON knockdown with treatment with BRAFi (Dabrafenib)—an inhibitor of B-raf, which is a factor stimulating cell growth by stimulating the MAP kinase/ERKs signaling pathway—caused a synergetic effect that reduced the tumor size (Fig. 12). Vice versa, recombinant expression of SAMMSON in the immortalized melanocytic cell line MEL-ST, which does not naturally express SAMMSON, conferred a growth advantage in vitro and in vivo (Fig. 13; Vendramin et al. 2018).
The overall idea and clinical importance
A major idea that Prof. Leucci is pursuing is that environmental cues (stress) can induce therapy resistance or metastasis at the molecular level through the regulation of lncRNAs. She showed convincingly indeed that cancers can select enhanced expression of certain lncRNAs to alleviate stress.
Very important is that Prof. Leucci found that downregulation of some lncRNAs makes cancer cells more sensitive to standard cancer therapies. Together with a company, she is aiming to test this in clinical trials.
Ongoing research
Prof. Leucci introduced to us a number of ongoing projects in her group, addressing a number of lncRNAs and various functions and diseases. Especially fascinating I found a study, which I can’t present here because she only just submitted it, that identified an lncRNA upregulated in different drug-escape variants of a cancer and if this lncRNA was knocked down the cancer cells became therapy-sensitive.
I must say that at some point in the past I have been somewhat skeptical about this line of research (because it incorporates so many variables and concepts), but the more I am diving into it (including the writing of this summary), the more I am getting convinced about its value and clinical importance. Judging from the pictures that Prof. Leucci showed us of her staff members, they all seem to share her enthusiasm, and it must be great to work in this field.
