EDITED: I think I completely misunderstood your question. And I gave a utterly irrelevant answer before, but I will leave it since it has been voted on. My apologies if I treated your question too simplistically !! Maybe I can make more sense here:

A. I think your first order is to determine - as you say - the lentiviral integration sites. This cannot be avoided. If you want to follow the single cell-single clone hypothesis, you have to show they have the exact same viral integration site.

B. Since you infected a population of cells, and each cell can semi-randomly incorporate a viral genome at a different site, you are back at step A where you have to follow cells by identity of integration site. If the sites differ that is in itself a biological variability for heterogenous responses.

C. A single integration event even in a a single cell can itself give rise to clones. It is well known that once an oncogenic event is initated - we assume that your lentivirus integrated into a tumor supressor or so - chromatin destabilization following mutation can lead to dramatic levels of somatic mutations which can then be selected upon. But all these clones with differing genomic mutations would share the same integration event of the parent.

Maybe this can help:

"Linear Amplification Mediated PCR for the retrieval of lentiviral vector integration sites from tumors induced by insertional mutagenesis."

Protocol Exchange

(2013)

doi:10.1038/protex.2013.009

Published online

30 January 2013

and : "Retroviral DNA Integration: ASLV, HIV, and MLV Show Distinct Target Site Preferences " PLOS Biology

http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.0020234

-----------------------------------------

[1. Form the figure it looks like you added all three viruses together is that right? Or were they added separately to separate populations of epithelial cells. I presume one would need to do it individually to understand what happens first? Maybe I have you wrong !

Even if done singly, every cell is expected to have a independent non-identical integration event - lentiviral integration sites are not restricted to one site but occurs at multiple sites although not indiscriminately as originally thought. i.e one cell type, but many independent integration sites. A few of these will knock out say a tumor supressor in one cell but may activate a unrelated oncogene in another. Thus, you have heterogeniety.....

Thus, it follows in lentiviral knockdowns of favorite genes, technically one should choose a clone derived from a single cell population, so that they derive from the same viral integration site, and that variable is taken out of the equation.]

EDITED: I think I completely misunderstood your question. And I gave a utterly irrelevant answer before, but I will leave it since it has been voted on. My apologies if I treated your question too simplistically !! Maybe I can make more sense here:

A. I think your first order is to determine - as you say - the lentiviral integration sites. This cannot be avoided. If you want to follow the single cell-single clone hypothesis, you have to show they have the exact same viral integration site.

B. Since you infected a population of cells, and each cell can semi-randomly incorporate a viral genome at a different site, you are back at step A where you have to follow cells by identity of integration site. If the sites differ that is in itself a biological variability for heterogenous responses.

C. A single integration event even in a a single cell can itself give rise to clones. It is well known that once an oncogenic event is initated - we assume that your lentivirus integrated into a tumor supressor or so - chromatin destabilization following mutation can lead to dramatic levels of somatic mutations which can then be selected upon. But all these clones with differing genomic mutations would share the same integration event of the parent.

Maybe this can help:

"Linear Amplification Mediated PCR for the retrieval of lentiviral vector integration sites from tumors induced by insertional mutagenesis."

Protocol Exchange

(2013)

doi:10.1038/protex.2013.009

Published online

30 January 2013

and : "Retroviral DNA Integration: ASLV, HIV, and MLV Show Distinct Target Site Preferences " PLOS Biology

http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.0020234

-----------------------------------------

[1. Form the figure it looks like you added all three viruses together is that right? Or were they added separately to separate populations of epithelial cells. I presume one would need to do it individually to understand what happens first? Maybe I have you wrong !

Even if done singly, every cell is expected to have a independent non-identical integration event - lentiviral integration sites are not restricted to one site but occurs at multiple sites although not indiscriminately as originally thought. i.e one cell type, but many independent integration sites. A few of these will knock out say a tumor supressor in one cell but may activate a unrelated oncogene in another. Thus, you have heterogeniety.....

Thus, it follows in lentiviral knockdowns of favorite genes, technically one should choose a clone derived from a single cell population, so that they derive from the same viral integration site, and that variable is taken out of the equation.]

Here is a user friendly review. There are many good ones like this.

This is a very good idea. We do integration site sequencing routinely using the nrLAM-PCR method. Nature Protocols had a paper on it in 2010, Paruzynski et al.

Monoclonal or polyclonal origin of tumour can be assessed successfuly by using cytogentic tools. Studying chromosomal aberrations pattern can guid you the original stem cell. If your result shows common rearrangement theme then you have monoclonal origin,if not then you may expect polyclonal origin.

I think the paper from Bob Kerbel's lab published in 1988 would speak to your question. Genetic tagging of tumor cells with retrovirus vectors: clonal analysis of tumor growth and metastasis in vivo. Korczak B, Robson IB, Lamarche C, Bernstein A, Kerbel RS. Mol Cell Biol. 1988 Aug;8(8):3143-9.

you can use a specific piece of DNA of your virus origin as a probe to do a genomic southern- on tumor DNA from different mouses.If you have the same pattern of bands in different tumors, its an indication that your retroviruses integrate into one same site of your human genome and cause a clonal tumor in all mice.

If you want to know if the tumor was from one cell origin, then you will need to do a serial dilution for transplantation, such as 1000 cells, 100 cells, 10 cells and 1 cell transplantation.

Hi All

Thanks for the useful feedback. We've tossed around a number of ideas to tackle this (Single cell sorting followed by integration analysis, Southern blotting, Limiting Dilution Analysis, Digital PCR, Laser Capture Microdissection, inverse-PCR, LM-PCR and LAM-PCR to name a few), however LAM-PCR was where we were thinking of heading (followed by Next Gen sequencing) due to the sensitivity.

We still have two main technical difficulties however.

1. Firstly, as Prem duly noted, I need to identify the insertion sites. LAM-PCR should suffice for this. Following this I then want to know which virus has inserted in what site. Here i am up against it with regards to read length. I say this as I'm infecting with multiple viruses simultaneously (up to 5 distinct lentiviruses), and the sequences that allow me to discriminate between the different viruses are (in some cases) 2kb from the junction of viral and cellular DNA (grrr... I should've used barcoded lentis!). I don't really have a solution to this. I have access to a PACBIO sequencer which could bridge this read issue with it's super long read length, but wonder if there is a better way out there (perhaps using several rounds of biotin-conjugated probes). If anyone has any thoughts regarding this I'd be most appreciative.

2. My second worry is whether this approach will tell me what I want to know, namely: has a given tumour arisen from a single infected cell? Will I be able to tell the difference between a clonal tumour and one where two infected cells gave rise to clones of equal size?

Thanks all for your input once again, it is very much appreciated.

Al

Can you make more tumors? If so, it may be nice to use the RGB marking method described by Boris Fehse's lab (Weber et al. Nature Medicine 2011, Nature Protocols 2012). This way, you can see the clonality, much the same way as Brainbow works.

WIth respect to knowing if something is clonal, you can do copy number analysis for the vector. If you detect four integrations and are fairly confident that there are no more, and if you have a vector copy number of four, you can be somewhat confident that you have a clonal tumor.

LAM-PCR and Next-gen sequencing - if you can afford it !!! - is the best way to go. All other methods are too labor intensive for the info you get. I don't know what the statistical considerations are for coverage since you have to deconvolve 5 lentiviruses (not a big deal I would guess !!!! :-) ). BTW, do not sweat the barcoding stuff. In my humble opinion, the whole point of next-gen sequencing is it supercedes barcoding.

I still think the lineage problem is not a problem, and can be resolved by comparison with the ORIGINAL population before implantation. If you go via the next-gen sequencing route, there are two 'markers' you could follow. The type and site of virus integrated, and the somatic mutations.

1. Irrespective of what somatic mutations have arisen, the daughter clones will be identifiable by the lentiviruses they have integrated. If a population of cells all share the same three - arbitrary number - lentiviruses integrated in the same sites the must be derived from common ancestors.

2. Even if two cells integrate the same three viruses, AND the exact same integration site (a low probability itself ??), I do not think they will share the exact same SOMATIC hypermutations. The somatic mutation info you get from next-gen info is very useful. These somatic mutations will propogate in a lineage fashion. When you overlap lentiviral integration site info and lentiviral type, I think a lineage will be obvious ....?

Make sense? Or have I missed something.......

Thanks once again all, I think I have all the answers I need. I'll be sure to let you know how I get on with this.