This paper was enjoyable to read and gave us space to design a model from their findings. Although the model may not have been good, it was a great practice to combine small pieces of information to figure out the big picture. My group focused on Question 4 – that lncRNA HOTTIP associates with CTCF protein. This was also good background information for my group’s final project when talking about future directions to study the function of a lncRNA.

Small cooperative assignment: Wang et al. paper

Assigned paper: Wang, F., Tang, Z., Shao, H., Guo, J., Tan, T., Dong, Y., Lin, L. (2018). Long noncoding RNA HOTTIP cooperates with CCCTC-binding factor to coordinate HOXA gene expression. Biochemical and Biophysical Research Communications 500 (2018) 852e859

A note about HOTAIR (a very famous line RNA involved in hox regulation): We will see in detail how another lncRNA, air, carries out the same function, but at an imprinted locus. So, don’t worry too much about it twhen it is discussed in the introduction.

Question 1 (everyone)

Before this paper, what was known to the authors about HOTTIP, what it does, and how it drives methylation of H3K4? What is the consequence of having a lot of H3K4me at a locus?

HOTTIP is a lincRNA, found in the HOXA locus.

Question 2 (Erin, Melissa, Elizabeth, Miguel, Cathy, Oline, Jasia)

Consider Figure 1.

1. What was the experiment that lead to those results? (What did the authors do, what did they measure)?

• • Used published data of:
    • ChIP-seq with CTCF, H3K4me3, and H3K27me3 at the HoxA locus
    • RNA-seq of HoxA (A1 – A13)

1. What question were the authors addressing in the experiment?

• What role does CTCF have on transcriptional control in Hox genes through a model of chromatin structure organization?
• Does the amount of histone modification (H3K27me3 and H3K4me3) correlate to RNA transcripts of the HoxA genes?
  • (edited from: the location of the HoxA genes? )
• What is the spatial relationship between CTCF binding sites to HoxA genes?

1. Why did the authors (most likely) chose this two particular cell lines?

• BJ and HFF cells are distal foreskin types of fibroblast cell lines and expression of HOTTIP was mainly observed in human fibroblasts and are not fully differentiated yet.
• They are distal, being from foreskin and hoxA genes are expected to be expressed in distal stages

1. In terms of the ChIP experiments, what additional control could the authors have included, and what would its purpose have been? [If you have trouble with this question, set it aside and come back to it at the end].

• Measured the amount of histone modifications in other cell lines, not fibroblast lines, as a negative control
• Measured the amount of different types of histone modifications in HoxA genes to show that specific histone modifications are involved in the expression of the HoxA genes
• For C and D, they can also test the CTCF antibody specificity at other DNA sites (other than the C1-C6 regions) that we believe do not associate with CTCF

1. What do the data show?

• Hox A transcripts are present at higher levels from regions that are also enriched for H3K4me3, which is typically associated with transcriptionally active regions
• CTCF is associated in  regions C1-6

1. What can we directly conclude from the data, and why?

• There is a direct relationship between H3K4me3 levels and levels of HoxA RNA transcript
• There is an inverse relationship between the H3K27me3 levels and levels of HoxA RNA transcript

1. In one sentence, what did the authors demonstrate?

• The authors conducted an observation experiment to lay out groundwork on putative CTCF binding sites and demonstrated a correlation between histone modifications enrichment and expression of hoxa cluster genes.

Question 3 (Tanis, Shannah, Meilin, Pareesa, Dee, Maria S, Amanda)

Consider Figure 2.

1. What were the experiments that lead to those results? (What did the authors do, what did they measure)?  

Knock out C5 (CRISPR) on HOXA gene cluster. They measured CTCF binding to C1-C6 domains on HOXA gene in BJ cells/HFF cell lines using ChIP assays. They also measured mRNA expression of HOXA genes when they knocked out C5 and C5 and HOTTIP knockout (RNAI with siRNA) and compared them to a wild type control.

 

• What questions were the authors addressing in the experiments?

 

Is there a functional connection between HOTTIP lincRNA and CTCF binding factors? If there is, how does it contribute to gene activation and chromatin organization? Do we observe effects on HOXA gene expression when we disrupt CTCF binding (via C5 region deletion on HoxA)  and knockdown HOTTIP respectively (via RNAi)?

Panel C: Is there differential HOTTIP RNA expression between BJ and HFF cells?/ Is there HOTTIP expression in living cells?

 

 

• The authors did a HOTTIP knock-down (siRNA) and not a knockout of the HOTTIP locus. Please provide two reasons why a knockout would not have been appropriate.

 

1. Knock-down siRNA doesn’t disrupt the HOXA gene, as only lincRNA is affected. If they did a knockout by removing HOTTIP, it would not be clear whether the observed effect was due to the knockdown itself or the disruption of the HOXA gene itself.
2. SiRNAs are extremely specific. The nucleotide sequence will specifically match the HOTTIP lincRNA. In comparison a knockout is much less specific, which could potentially lead to unintentional effects.

 

• What is the purpose of running the experiments on two different cell lines?

 

To ensure that the observed effects are not specific to the cell line.

 

• What do the data show?

 

A+C – Deletion of CTCF core motif leads to abrogation of CTCF occupancy at C5 site in both BJ and HFF cells. Sites C4 and C6 are also affected and there is a large decrease in CTCF binding.

B+D – Deletion of C5 site leads to upregulation of HOXA6 (gene immediately rostral of C4 and C5 sites) in comparison to WT control. HOXA3-5 have increased expression whereas HOXA1 and HOXA2 are fully repressed.

When HOTTIP was knocked down, this abrogated the upregulation effect of HOXA3-6 genes after C5 deletion.

 

• Are any of the data surprising? If so, which parts?

 

There was a dramatic decrease in CTCF binding at the neighbouring sites C4 and C6.

 

• What can we directly conclude from the data, and why?

 

We can directly conclude:

• C5 is necessary to bind CTCF to HOXA gene clusters, at C5 and also is necessary for maximal C4 and C6 to CTCF.

1. In one sentence, what did the authors demonstrate?

CTCF and HOTTIP coordinate together to sho

 

Question 4 (Heather, Ana-Maria, Brett, Jamie, Jacob, Josh, Beth)

Consider Figure 3.

1. What was the experiment that lead to those results? (What did the authors do, what did they measure)?
   1. The experiment used a CRISPR-mediated gene KO of the ∆C5 CTCF binding element within the HoxA DNA cluster. Histone methylation marks H3K27me3 and H3K4me3 were measured using ChIP-Seq against two different human foreskin fibroblast cell lines, BJ and HFF. By this method, the authors were able to determine the presence and quantity of the two histone methylation marks at the various HoxA loci, A1-A13. As a final manipulation, the authors also used an siRNA-mediated knockdown of the HOTTIP lncRNA, followed by ChIP-seq, to study the effects of HOTTIP on HoxA locus histone methylation
2. What question were the authors addressing in the experiment?
   1. The authors wanted to investigate how HOTTIP and CTCF are related to HoxA gene expression levels through chromatin modifications (methylation of histone subunits).
3. What additional ChIP control could the authors have included, and what would its purpose have been? [If you have trouble with this question, set it aside and come back to it at the end].
4. The authors did a HOTTIP knock-down (siRNA) and not a knockout of the HOTTIP locus. Please provide two reasons why a knockout would not have been appropriate.
   1. Because Hox genes are expressed in a collinear manner; i.e. they are expressed in a particular order along the body axis during development, a knockout of the HOTTIP locus may result in disruption of the spatial regulation and the HoxA cluster, whereas siRNA would have no effect on the spatial location of the locus.  
   2. In addition, an RNA knock-down allows temporal control of the experiment, so that the siRNA can be reversed if necessary
   3. Ensures no disruption of the hox cluster structurally or sequence wise
5. What do the data show?
   1. In BJ cells, the H3K4me3 mark across loci A13-A7 were almost identical (capturing ~80% of the input) in terms of presence and quantity between WT and ∆C5 cells; siHOTTIP + ∆C5 cells had between 20 and 40% of the total input capture. Significantly reduced H3K4me3 was seen in WT cells across the A6-A2 locus relative to ∆C5 cells (WT cells captured between 5-20% of the input), and even presence/quantity of the mark across all conditions at the A1 locus.
   2. Also in BJ cells, the H3K27me3 mark presence across loci A1, A2, and A7-13 between the WT, ∆C5 and ∆C5+siHOTTIP cells were fairly identical, though the percentage of input captured varied, with there being a fairly constant gradient from the lowest input covered in A13-10 (~8%) to the highest input covered (~50%) in A1. However, the WT percent capture in A3-6 was on average 2 fold greater than that of the ∆C5 and ∆C5+siHOTTIP knockdowns, hovering around 50% whereas the ∆C5 and ∆C5+siHOTTIP knockdowns decreased over the A3-A6 gradient, from ~30% in A3 to ~20% in A6. This indicates significantly reduced H3K27me3 in those loci in the knockdown experimental groups.
   3. In HFF cells, there is no significant difference in
6. Are any of the data surprising? If so, which parts?
   1. Functional lincRNAs are not well characterised, and so although they may be abundant, it is still a surprising/rare phenomenon to come across in gene regulation experiments at the moment
7. What can we directly conclude from the data, and why?
   1. The C5 binding site on the HoxA locus is required for the normal (WT) H3K4me3 (“activating”) and H3K27me3 (“silencing”)  histone modifications levels respective to the HoxA genes. The addition of an siRNA targeting the HOTTIP RNA showed that HOTTIP is necessary for the H3K4me3 histone modifications on the HoxA genes, but not for the H3K27me3 histone modifications in both BJ and HFF cells.
8. In one sentence, what did the authors demonstrate?
   1. The authors demonstrated that HOTTIP in coordination with CTCF binding  is required for maintenance of epigenetic modifications that regulate expression of the HoxA genes in foreskin fibroblasts.

 

Question 5 (Kate, Maria V., Evan G., Joanne, Evan Z, Hayden)

Consider Figure 4.

 

• What were the experiments that lead to those results? (What did the authors do, what did they measure)? You may have to separate out the different parts of the figure (different experiments).

 

PANEL A: The authors engineered E. coli cells to express recombinant GST and GST-tagged CTCF. They purified these proteins, and ran them on a gel. Panel A shows this gel, to demonstrate that they successfully purified these proteins.

 

PANEL B: The purified GST and GST-CTCF were incubated with HOTTIP RNA, or control histone mRNA. Any RNA not bound to these purified proteins was washed away. RT-PCR was used to show whether either protein was able to pull down HOTTIP RNA, as compared to the histone control which is known to bind to neither protein.

 

PANEL C: This shows the results of an IP experiment, pulling down the CTCF protein from cultured human foreskin fibroblasts. The authors then use RT-PCR to measure the amount of HOTTIP RNA was found to be associated with their immunoprecipitated CTCF protein, as compared with the negative control IP for IgG.

 

PANEL D: This is an in vivo experiment, wherein the authors use an anti-biotin antibody to pull down biotinylated HOTTIP RNA, along with GFP and antisense HOTTIP RNA to serve as negative controls. They separate any associated proteins, and run then on a gel. They probe for CTCF, WDR5 (a protein to which HOTTIP is known to associate) and a-tubulin (a protein to which HOTTIP is known to have no association).

 

 

• What question were the authors addressing in each of the experiments?

 

Figure 4A – Have we successfully purified the recombinant proteins that we will be using to test interaction between HOTTIP and CTCF

Figure 4B- To what levels do HOTTIP RNA and GST-CTCF interact in comparison to GST?

Figure 4C- Is there differential HOTTIP RNA expression between BJ and HFF cells?/ Is there HOTTIP expression in living cells?

Figure 4D – Does HOTTIP bind to CTCT? Does HOTTIP retrieve CTCF?

 

 

• What is the purpose of the histone mRNA in panel B

 

Histone mRNA is a control to show that the induced retrieval of HOTTIP RNA by GST-CTCF compared to GST is due to specific interactions of HOTTIP and GST-CTCF rather than random chance. Histone mRNA is a negative control in which we do not expect induced retrieval in either of the proteins. Because the levels of histone mRNA retrieved by GST and GST-CTCF are the same, we know that the experimental system is valid (?) and describe that HOTTIP RNA retrieval by HOTTIP RNA is higher.

 

 

• What do the data show?

 

Figure 4A – Western blot shows successful purification of GST-CTCF, which were used to test HOTTIP RNA binding/retrieval in vitro

Figure 4B – GST-CTCF and not GST (control) retrieved HOTTIP RNA and not histone mRNA (control), as measured by qRT-PCR

Figure 4C – HOTTIP RNA is co-precipitated with CTCF but not with IgG IP (control) in both BJ and HFF cells

Figure 4D – HOTTIP RNA, and not antisense HOTTIP RNA or GFP, retrieves CTCF in vivo, and WDR5 (positive control), but does not retrieve alpha-tubulin (negative control)

 

 

• What can we directly conclude from the data, and why?

 

We can directly conclude that HOTTIP RNA directly interacts with CTCF in vivo and in vitro.

1. What would be your next experiment, if you were one of the authors?

HOTTIP is a lncRNA with 3764 nucleotides, so I would attempt to identify the region of HOTTIP that interacts with CTCF. To do this I would amplify HOTTIP RNA, shear it into smaller fragments of RNA, use CTCF to pull out RNA from this sample, and sequence the resulting RNA. From there, I could build a consensus sequence from the various sequenced fragments. This consensus sequence would likely indicate the specific region within HOTTIP that interacts with CTCF.

 

• In one sentence, what did the authors demonstrate?

 

• Results reveal that there is a functional connection between HOTTIP and CTCF, and sheds light on lincRNAs (long intergenic non coding RNA) involvement in gene activation.