23 replies to this topic

[gyma]

Hi guys. i asked a question (http://www.protocol-...showtopic=13875) a week ago and i got very valuable comments, thank you very much. now i have a problem which never happened before and i cannot understand.
i used 2 primers to amplify a 2kb region in proviral genome which is integrate to carriers. at first i did this in a virus-transfromed cell line, sequenced the product and use it as the reference sequence. then i used the same condition to amplify this region in carriers. it is very strange that i got 4 bands for every carrier, instead of 1 in cell line. the size ranged from 0.8kb to 4kb. at first, i thought those might be nonspecific bands so i just cut out the 2kb band and cloned it into TOPO vector and sequenced it. i got the right sequence for the first 2 carriers i sequenced. then i optimized the insert/vector ratio in the ligation and i found i could use much less pcr product and get not less clones than before. however, when i sequence these clones, i got no result except from vector-based primers. and more surprisingly, the sequence is totally different from proviral genome that i had sequenced in the first 2 carriers. i blasted it and found it has 100% homology with a region in chromosome 13q. i dont understand. the only difference in method is the insert/vector ratio. is that the reason? or its the problem of different carriers? why cant i find even the primer sequence?
i dont konw how did this happen. trying to find some possible reason, i sequence the other 3 bands and found they all localized in chromosome. it is very strange because when i use the primer sequence to do a homolog search in the sequence, i got no homolog at all. what are these fragments? its driving me crazy, please help me. thanks.
i attached the photo, 2nd band from top is what i need, 2kb.

Attached Thumbnails

• 

Edited by gyma, 30 March 2010 - 04:26 AM.

[merlav]

First take the primer sequence and check with Blast.  If the tool recognize the sequence for the virus and not for human the problem could be too low temperature of annealing .  rise the temp until you see ONE band of the size expected, don't clone or sequence until you got the band.  If the Blast tool recognize the sequence for both, then the design of the primers are bad. The best solution throw away the primers and begin for design a new primers set(is cheaper than continue cloning and sequencing with no results).  I know that sometimes we think that the design was good, but sometimes is not, so we have to let it go the bad primers....It doesn't matter who design it.  Even the ones that are cited in a paper can be wrong and must be check first.I can tell this by experience. Use primer3, primer express or other primer design tool.  Try to get a set with a high Tm (60C or higher), low repeats, low secondary structures, etc.    
Good luck!!!

[HomeBrew]

gyma, on Mar 30 2010, 05:39 AM, said:

however, when i sequence these clones, i got no result except from vector-based primers. and more surprisingly, the sequence is totally different from proviral genome that i had sequenced in the first 2 carriers. i blasted it and found it has 100% homology with a region in chromosome 13q. i dont understand. the only difference in method is the insert/vector ratio. is that the reason? or its the problem of different carriers? why cant i find even the primer sequence?

The fact that your insert-based primers produce no product, and that your sequenced insert includes no primer sequences means that you didn't clone a PCR product -- you cloned a genomic fragment.  Thus, your problem here is not related to your primers; they played no role in generating what you cloned...

How do you prepare your template DNA and your vector DNA?

[gyma]

thank you very much, merlav and homebrew.
i think i have found the problem. it is the primer, as merlav had said. this primer set worked well in the cell line. and not bad for the first 2 carriers i have sequenced. because i got over 15 clones with the right sequence among 20 picked. the other clones also have a 2kb insert but its from human genome, instead of proviral genome. the problem is this primer set doesnt work anymore, for the rest carriers. i think the amount of unspecific 2kb product is far more than the specific one. so i picked 20 clones and got nothing.
now i already designed a new primer set. i selected regions about 50bp from before and extended primer length from 20 to 25. Tm of these primers are around 60c. I will try again when the primers arrive.
thank you very much.

[gyma]

HomeBrew, on Mar 30 2010, 09:53 PM, said:

gyma, on Mar 30 2010, 05:39 AM, said:

however, when i sequence these clones, i got no result except from vector-based primers. and more surprisingly, the sequence is totally different from proviral genome that i had sequenced in the first 2 carriers. i blasted it and found it has 100% homology with a region in chromosome 13q. i dont understand. the only difference in method is the insert/vector ratio. is that the reason? or its the problem of different carriers? why cant i find even the primer sequence?

The fact that your insert-based primers produce no product, and that your sequenced insert includes no primer sequences means that you didn't clone a PCR product -- you cloned a genomic fragment.  Thus, your problem here is not related to your primers; they played no role in generating what you cloned...

How do you prepare your template DNA and your vector DNA?

homebrew, you are right. it seems that what i gel purified is a 2kb genomic dna. it may also contain a rare amount of proviral genomic product. that is why i cant get right, positive clones.
my template DNA is genomic DNA extracted from PBMC of carriers. about vector DNA, i used TOPO vector from invitrogen, which is for blunt end ligation. thank you for your reply.

[HomeBrew]

Okay, I *think* this is making sense now...

When you got the human DNA insert, did you just sequence one clone? Or a few clones from just one experiment, such that the clones you sequenced could have been siblings?

If you run out a sample of your carrier genomic DNA, does it appear with the same discrete banding pattern?

What I'm thinking is that both merlav and I have a piece correct -- your primer set is not working for whatever reason (one or both of your primers has degraded or one or both of them don't recognize anything in the carrier DNA, despite the fact that they did in the transformed cell line).  Additionally, your carrier DNA is being fragmented by some mechanism (perhaps randomly, perhaps enzymatically) into blunt-ended pieces, and thus these fragments can be successfully cloned into your blunt-ended vector.

I suspect if you did the same experiment again, you'd recover a different piece of the carrier genome...

So, I think you definitely need new primers -- either re-synthesize the ones you've used, or redesign a set that will recognize what you're looking for in your carrier DNA.  I also think you should either try to block whatever is producing the fragmentation of your carrier DNA (if it's enzymatic) or redesign your approach so you avoid blunt-end ligation.

Can you attach restriction enzyme sites to your primers, and use a digested vector to allow sticky-end ligation?

[gyma]

HomeBrew, on Mar 31 2010, 09:33 PM, said:

Okay, I *think* this is making sense now...

When you got the human DNA insert, did you just sequence one clone? Or a few clones from just one experiment, such that the clones you sequenced could have been siblings?

If you run out a sample of your carrier genomic DNA, does it appear with the same discrete banding pattern?

What I'm thinking is that both merlav and I have a piece correct -- your primer set is not working for whatever reason (one or both of your primers has degraded or one or both of them don't recognize anything in the carrier DNA, despite the fact that they did in the transformed cell line).  Additionally, your carrier DNA is being fragmented by some mechanism (perhaps randomly, perhaps enzymatically) into blunt-ended pieces, and thus these fragments can be successfully cloned into your blunt-ended vector.

I suspect if you did the same experiment again, you'd recover a different piece of the carrier genome...

So, I think you definitely need new primers -- either re-synthesize the ones you've used, or redesign a set that will recognize what you're looking for in your carrier DNA.  I also think you should either try to block whatever is producing the fragmentation of your carrier DNA (if it's enzymatic) or redesign your approach so you avoid blunt-end ligation.

Can you attach restriction enzyme sites to your primers, and use a digested vector to allow sticky-end ligation?

thanks for these valuable suggestions.
Maybe I didnot make it very clear before. I used high-fidelity polymerase so the PCR product is supposed to be blunt-ended. so I think fragmentation wouldnot be the problem.
I considered it again and think it would be no problem if I had ever set the negative control, which should be genomic DNA of PBMC from virus negative people. If I got same band pattern from negative control, then its definitely the problem of primers.
yesterday I also tried the rest of primer sets (20bp) for other regions of proviral DNA. same things happened again. they worked well in the cell line, but showed either a confusing pattern or no band at all in one carrier that I tested. Now I started to suspect this method.
anyway, I am going to test the new primer set (25bp) today. If extending primer length works, I would also redesign other primer sets. but if it doesnt work, then......I need to think...

[HomeBrew]

gyma, on Mar 31 2010, 09:45 PM, said:

I used high-fidelity polymerase so the PCR product is supposed to be blunt-ended.

Yes, but if you attach restriction enzyme sites to the 5' end of your primers, the PCR product (still produced using high-fidelity polymerase) can be digested and cloned using sticky-end ligation.

Another question arises -- if, as it seems, the spurious bands you're seeing on your gels are fragments of your genomic template, why are you seeing them?  A negative PCR should be negative -- there's no way you should be seeing non-amplified template bands as bright as those in your photo -- how much template DNA are you using?

[gyma]

HomeBrew, on Apr 1 2010, 07:51 PM, said:

gyma, on Mar 31 2010, 09:45 PM, said:

I used high-fidelity polymerase so the PCR product is supposed to be blunt-ended.

Yes, but if you attach restriction enzyme sites to the 5' end of your primers, the PCR product (still produced using high-fidelity polymerase) can be digested and cloned using sticky-end ligation.

sticky-end ligation should have higher efficiency than blunt-end, but that requires extra steps like digest, purification and may cause loss of inserts. so I used TOPO for convenience. and now after optimization, the efficiency wont be a problem.

Another question arises -- if, as it seems, the spurious bands you're seeing on your gels are fragments of your genomic template, why are you seeing them?  A negative PCR should be negative -- there's no way you should be seeing non-amplified template bands as bright as those in your photo -- how much template DNA are you using?

I thought since the nonspecific bands are amplified from human genome, instead of proviral genome, if I run PCR in proviral genome-negative sample DNA, I should be able to get at least very similar band pattern. because the specific product has a very small portion. I used normally 100ng genomic DNA in a 20ul reaction for cell line and 50~400ng for carriers. If I only do it for a test, I will use 50ng, and if I want to purify the product and sequence it, I will use more.

Thank you very much, Homebrew.
I tested the 25bp long primer set, it doesnt work. now I am not sure what to do next. but I have one thought now. the only difference between the correct band and the wrong band is the sequence, the wrong band doesnt have even the primer sequence. maybe I could use a primer set with enzyme sites so after PCR I will digest the product and do sticky ligation. but what if the wrong product had the same enzyme site?  

Attached Thumbnails

• 

Edited by gyma, 01 April 2010 - 04:30 AM.

[HomeBrew]

gyma, on Apr 1 2010, 08:06 AM, said:

I thought since the nonspecific bands are amplified from human genome...

Yes, but your evidence suggests these bands are *not* amplification products -- if they were, they would (by definition) have the primer sequences in them.  Thus, if they're not amplicons, why are you seeing them?  If the amount of template you're using is sufficient to produce fragment bands that bright on an EtBr-stained gel, you're using way too much template, and your PCR reaction would likely fail, even with perfect primers...

gyma, on Apr 1 2010, 08:06 AM, said:

...but what if the wrong product had the same enzyme site?  

Exactly why we must first investigate the source of all that banding...

[phage434]

Step 1:  Create a PCR reaction with twice your normal volume.  Split it in half.  Cycle half.  Run a gel with a marker lane, uncycled and cycled reactions in adjacent wells.

Your uncycled lane should show no bands whatsoever, because it should have sufficiently low template concentration that it is not visible.

[gyma]

HomeBrew, on Apr 2 2010, 02:12 AM, said:

gyma, on Apr 1 2010, 08:06 AM, said:

I thought since the nonspecific bands are amplified from human genome...

Yes, but your evidence suggests these bands are *not* amplification products -- if they were, they would (by definition) have the primer sequences in them.  Thus, if they're not amplicons, why are you seeing them?  If the amount of template you're using is sufficient to produce fragment bands that bright on an EtBr-stained gel, you're using way too much template, and your PCR reaction would likely fail, even with perfect primers...

gyma, on Apr 1 2010, 08:06 AM, said:

...but what if the wrong product had the same enzyme site?  

Exactly why we must first investigate the source of all that banding...

Well, if those bands are not amplified by PCR, then why increasing cylce number resulted in increased amount? I attached a photo showing cycle number 30,35,40 from left to right in 1 PCR reaction of 1 carrier. I think you could see the brightness of all 4 bands increases as the cycle number increases. I am sure I didnt use too much template, 50 ng isnt too much for a PCR reaction, right?

Attached Thumbnails

• 

Edited by gyma, 01 April 2010 - 07:09 PM.

[gyma]

phage434, on Apr 2 2010, 03:20 AM, said:

Step 1:  Create a PCR reaction with twice your normal volume.  Split it in half.  Cycle half.  Run a gel with a marker lane, uncycled and cycled reactions in adjacent wells.

Your uncycled lane should show no bands whatsoever, because it should have sufficiently low template concentration that it is not visible.

do you mean the bands i got from PCR already exist in the template?

[HomeBrew]

We're trying to figure out if those bands are amplicons.  phage434's suggestion is designed to run two identical samples -- each comprising one-half of a PCR reaction mix that was made with twice the volume of everything you usually use in a mix, and then split -- thus the only difference between them is that one mixture was subjected to PCR cycling, and the other wasn't.  If the bands are present and visible on a gel in the sample that wasn't cycled, they're not amplicons.

The whole thing comes down to the fact that your primer sequences were not in the insert you sequenced, meaning that that particular insert was not the result of a PCR amplification, and thus was a genomic fragment.  What we really need is an answer to the question I posed way back here -- did you just sequence one or a few clones from a single transformation?

If so, you may have just accidentally recovered a genomic fragment of the appropriate size that was mixed in with a population of amplicons of the same size.  Since you're blunt-end cloning, any fragment in a population of same-sized fragments has an equal chance of being cloned, regardless of whether it was generated by PCR or it was just a random genomic fragment that migrated with the amplicons.

[gyma]

HomeBrew, on Apr 2 2010, 01:10 PM, said:

We're trying to figure out if those bands are amplicons.  phage434's suggestion is designed to run two identical samples -- each comprising one-half of a PCR reaction mix that was made with twice the volume of everything you usually use in a mix, and then split -- thus the only difference between them is that one mixture was subjected to PCR cycling, and the other wasn't.  If the bands are present and visible on a gel in the sample that wasn't cycled, they're not amplicons.

The whole thing comes down to the fact that your primer sequences were not in the insert you sequenced, meaning that that particular insert was not the result of a PCR amplification, and thus was a genomic fragment.  What we really need is an answer to the question I posed way back here -- did you just sequence one or a few clones from a single transformation?

If so, you may have just accidentally recovered a genomic fragment of the appropriate size that was mixed in with a population of amplicons of the same size.  Since you're blunt-end cloning, any fragment in a population of same-sized fragments has an equal chance of being cloned, regardless of whether it was generated by PCR or it was just a random genomic fragment that migrated with the amplicons.

I just finished the PCR with a normal control from PBMC of a healthy person. although the normal control showed similar band patterns, it is quite different in the 2nd band, which is cloned and sequenced. if the 2nd band is specific, why does it contain human genome sequence?

Attached Thumbnails

• 

Edited by gyma, 02 April 2010 - 12:40 AM.