Hi Kate,

T-RFLP determines the structure of the dominant microorganisms in your samples. The microbial community structure in T-RFLP relates to the number and the relative abundance of the different fragments (T-RF). So when you perform T-RFLP, you want to ensure that you have a relevant number of T-RF to accurately determine your microbial community structure (this also applies to any/most techniques that determine microbial structure or diversity). In other words, you want the resolution of the T-RFLP to be high to determine accurately the microbial community structure. The logic is that: more you have T-RF, better is your resolution. Obviously you do not want to have too little T-RF, otherwise you may miss important information to determine microbial community structure. On the other hand, there is a limit to which having more T-RF will not had much information (and this limit come quite quickly with T-RFLP). If you are asking me what is the best number of T-RF, I will say it depends on the type of microorganisms and sample you work on, but a good T-RFLP profile (after cleaning T-RF for background noise and other stuff) is ~40 T-RF (for soil bacteria, one restriction enzyme used). I insist, I give this number but it really depends on what you are working on, and how you do your T-RFLP, this is just a guideline but not an absolute value or a target.

You can increase the number of T-RF, and potentially the resolution of your T-RFLP by either labelling both primer with different dye, or using multiple restriction enzymes. The restriction enzyme will cleave the DNA you amplified at specific site. However, those sites will not be the same from the reverse or the forward primer. So for the same sequence, you could for example have a T-RF of 150 bp from the forward primer, but only 30 bp from the reverse primer. So, the idea behind labelling both primer with different dye, is to increase the probability for your sequences to be cleave and produce a correct T-RF, as if it does not work properly from one side (i.e. to short or too long T-RF), it might work from the other side. On the other hand, this represent a major problem, because for a specific sequence if the enzyme cleave the sequence you always end up with 2 T-RF (nearly 100% sure to be of different size) and you artificially increasing the number of T-RF in your sample. This is the reason to use 2 different dyes for each primer. So you can analyse separately the T-RF from the forward or reverse primer to avoid this issue. This way, you can see if you get the same microbial community structure with forward and reverse primer, or if there is one side better (e.g. higher number of T-RF, results matching obvious expectation) than the other, to determine your community structure. Another aim of labelling both primers, and more important than trying to increase T-RF: is to duplicate your T-RFLP to determine the consistency of your profile/results (can you trust your results). You might have the same structure from both primer, and you state it in your article, and say that you will only focus on one primer. So, to answer your question: “how should I analyse my data”, I would analyse separately the data from forward and reverse primer to avoid this issue.

Different enzymes do not cleave at the same site. Restriction enzymes cleave at specific site, for example AluI cleaves between G and C if there is the following order of nucleotide AGCT, for HaeIII it is GGCC. So, if a restriction enzyme does not cleave a sequence because the 4 successive nucleotides are not present (or not at the right place, too short or too long T-RF), it might work for another one. So, that’s way you can increase the resolution of your T-RFLP, by potentially cleaving more sequences. Furthermore, some enzymes are not “good” for the amplicons you want to cleave, so you have to choose properly your enzyme(s) (main factor affecting T-RFLP results). The problem with using several restriction enzyme, is a bit similar than for dual labelling of primer (but not as bad) is that you can end up with 2 different T-RF for a specific sequence, because a specific sequence can have several restriction sites. So you can analyse your data separately and see if you get the same pattern/resolution but you could also merge your results from both enzymes (but only one primer) and do one analysis and see if you improve your data.       

To analyse your data, you analyse both primers separately. Then, you should probably try to analyse your data for one primer, either including both enzymes or for each enzyme. When you want to merge results from your 2 enzymes, what you can do is simply to copy and paste the data (after alignement) from one enzyme under the result from the other one. You could also merge before alignement but I will not go into details now why doing one or the other. I know it is extra work to do the analysis on the different combination, but you need to test if having 2 labelled primers, and multiple enzymes is useful or not and if you should keep going into that direction or if with 1 labelled primer and 1 enzyme, you already get the answer you want. I would expect that you do not need to label both primer, but using several enzymes is probably useful for AM fungi and you could either work on both enzymes separately or merge the results.  

All these information are often not properly stated in articles using T-RFLP, and it is confusing and ender the replicability of science…

Hope my reply will help you, and sorry for its length…

The reason to use 2 labelled primers is to get high number or enough T-RFs to characterise your microbial community. However, when working on bacterial or fungal community in soil, you do not need to label the 2 primers because you have enough T-RFs with 1 labelled primer to characterise your community accurately. Furthermore, having 2 labelled primers have some issues: not all the dyes are good to perform T-RFLP (see article from Pandey et al 2007), the signal from the 2 dyes might overlap each other and lead to problem identifying the origin of the signal. So, using one dye is easier and work well. The only situation I might advise to use 2, it is when you have only few T-RFs across all your samples (< 10 or 5 T-RFs), although it depends if you work on a system where you are expecting high or low microbial abundance/diversity. Even if you get low T-RFs number, I would probably advise to use more than one restriction enzyme rather than 2 dyes to get higher number of T-RFs as it is likely to give you more information/T-RFs.

Thank you very much for the clarification. I'm studying bacterial endophytes in seeds and roots. Diversity is expected to be lower in both roots and seeds compared to soil samples, though abundance of bacterial endophytes is unsure. I am also planning on using more than three restriction enzymes for my studies. 

I agree with Almeric, that it is better to use up to 3 restriction enymes, but you have to choose them well. Each enzyme should give you a number of TRFs for your particular sample. For actinobacterial endophytes please see the following papers

Conn, V. M. and Franco, C. M. M. (2004). Analysis of the endophytic actinobacterial population in the roots of wheat (Triticum aestivum L.) by Terminal Restriction Fragment Length Polymorphism (T-RFLP) and sequencing of 16S rRNA clones. Applied and Environmental Microbiology 70:1787-1794.

Conn, V.M., & C.M.M. Franco (2004). “The effect of microbial inoculants on the indigenous actinobacterial endophyte population present in the roots of wheat as determined by Terminal Restriction Fragment Length Polymorphism (T-RFLP).” Appl Environ Microbiol 70:6407-6413.

Thank you Sir Franco for the wonderful papers. I was able to get your 2004 article on analysis of Actinobacteria on roots of wheat long back and was actually thinking of doing a study on effects of inoculation, but was very thankful that you also showed your paper on effects of inoculants since my proposed study is quite similar except that stress is involved during inoculation. Aside from studying endophyte diversity, my practical objective is to look for effective inoculants compatible with both the plant host and its microbiota.

@Sir Almeric and Sir Franco, In relation to TRFLP analysis, I am just wondering if you encountered problems with amplification of plant based DNA especially plastids (chloroplast) and mitochondria. The use of specific primers (I would use 799f which excludes chloroplast DNA but not mitochondrial DNA) would be helpful but would not completely remove all the PCR-based complications. Primer dimers, chimera etc, could also occur and I am wondering how to identify them from the results of TRFLP.

Thank you once again.

First of all, those problems that you mentioned are not specific to T-RFLP but to any DNA/PCR-based methods. So the precautions you would take with any PCR are relevant for PCR that you will use for T-RFLP too.

Making sure you have specific primer is important. To ensure that your primer do not amplify non-desired sequence, you obviously run first your PCR product on an agarose gel (up to 2% if you want to properly separate potential bands of different sizes) to assess the specificity and size of your products. Then to really ensure you have specific amplification, you need to do some sequencing on your products to determine if you do not have unwanted sequences If you work with a simple system for which you know the richness, you can perform in silico digestion of sequences from database, or that you sequenced, to determine which size of T-RFs you are expecting.

Primer dimers are usually small in size, mainly limited to the size of the primer. So for T-RFLP when you analyse your results, you remove the T-RFs 500 bp, although this depends on the size of your PCR products and also the standard you use in your run.

Thank you for the suggestions Sir. I have 16S sequences of bacterial isolates from my seed samples which I could use for in silico digestion.

Most of the papers I have read have excluded 500, although I am perplexed on why and how a few authors have excluded specific T-RFs in between (50-500bp) simply stating that they are plant-based T-RFs without mentioning additional criteria for exclusion. I am only guessing that they used a data-based program for in silico identification of T-RFs.

There is no universal T-RFs related to plants that you could remove in any study. To determine if one of your T-RFs is from plant, you have to sequence your PCR product, and if you find sequences from plant you perform an in silico digestion to determine the size of the T-RF and then you can remove your fragment. However, you are likely to find the same T-RF related to different microorganisms or plant, and the fact of having some sequences originating from plant might be minor in comparison to bacteria for the same T-RF. Overall, T-RFLP is a robust technique and I would not worry too much unless the system you are working on is likely to contain mainly plant DNA and few “bacteria”.

Thank you once again for the valuable suggestions. I would try to produce data and results and update on these topics. Godspeed and all the best.

I got an article published online today in Biology and Fertility of Soils that discuss about T-RFLP: the effect of restriction enzymes selection, peak area threshold, matrix (relative abundance vs presence/absence), and statistical analysis on the biological conclusions. My article could be relevant for you at different levels. I hope you will find it useful.

Analysing the effect of soil organic matter on bacterial communities using T-RFLP fingerprinting: different methods, different stories?

A. Blaud, F. Diouf, A. M. Herrmann, T. Z. Lerch

http://link.springer.com/article/10.1007/s00374-015-1041-0

Thank you for sharing your article Sir. It would help me in many ways especially during analysis and discussion of the results I would be getting -- keeping my fingers crossed in every step of the way.

Hi,

Following on from your conversations about this topic I wonder if anyone could help me get my head around the use of two labelled primers and two restriction enzymes for TRFLP.If someone could explain in really basic terms what is happening and why that would be great. I am really struggling to visualise what is happening.

I am attempting to profile AM fungal communities from plant roots. All the papers I have read use a universal eukaryotic primer and an AM fungal primer. Both these are labelled with two different dyes. I then digest PCR products in separate restriction digest reactions, I am using two enzymes. Therefore I generate four variations of each sample; Primer1/enzyme1, Primer2/enzyme1, Primer1/enzyme2and Primer2/enzyme2.

I am confused as to why I need to do this and also I am then unsure how to proceed to analyse the data. How do I combine the four profiles to finish with a raw data file I can input to PRIMER and analyse.

Best wishes,

Kate

Hi Kate,

T-RFLP determines the structure of the dominant microorganisms in your samples. The microbial community structure in T-RFLP relates to the number and the relative abundance of the different fragments (T-RF). So when you perform T-RFLP, you want to ensure that you have a relevant number of T-RF to accurately determine your microbial community structure (this also applies to any/most techniques that determine microbial structure or diversity). In other words, you want the resolution of the T-RFLP to be high to determine accurately the microbial community structure. The logic is that: more you have T-RF, better is your resolution. Obviously you do not want to have too little T-RF, otherwise you may miss important information to determine microbial community structure. On the other hand, there is a limit to which having more T-RF will not had much information (and this limit come quite quickly with T-RFLP). If you are asking me what is the best number of T-RF, I will say it depends on the type of microorganisms and sample you work on, but a good T-RFLP profile (after cleaning T-RF for background noise and other stuff) is ~40 T-RF (for soil bacteria, one restriction enzyme used). I insist, I give this number but it really depends on what you are working on, and how you do your T-RFLP, this is just a guideline but not an absolute value or a target.

You can increase the number of T-RF, and potentially the resolution of your T-RFLP by either labelling both primer with different dye, or using multiple restriction enzymes. The restriction enzyme will cleave the DNA you amplified at specific site. However, those sites will not be the same from the reverse or the forward primer. So for the same sequence, you could for example have a T-RF of 150 bp from the forward primer, but only 30 bp from the reverse primer. So, the idea behind labelling both primer with different dye, is to increase the probability for your sequences to be cleave and produce a correct T-RF, as if it does not work properly from one side (i.e. to short or too long T-RF), it might work from the other side. On the other hand, this represent a major problem, because for a specific sequence if the enzyme cleave the sequence you always end up with 2 T-RF (nearly 100% sure to be of different size) and you artificially increasing the number of T-RF in your sample. This is the reason to use 2 different dyes for each primer. So you can analyse separately the T-RF from the forward or reverse primer to avoid this issue. This way, you can see if you get the same microbial community structure with forward and reverse primer, or if there is one side better (e.g. higher number of T-RF, results matching obvious expectation) than the other, to determine your community structure. Another aim of labelling both primers, and more important than trying to increase T-RF: is to duplicate your T-RFLP to determine the consistency of your profile/results (can you trust your results). You might have the same structure from both primer, and you state it in your article, and say that you will only focus on one primer. So, to answer your question: “how should I analyse my data”, I would analyse separately the data from forward and reverse primer to avoid this issue.

Different enzymes do not cleave at the same site. Restriction enzymes cleave at specific site, for example AluI cleaves between G and C if there is the following order of nucleotide AGCT, for HaeIII it is GGCC. So, if a restriction enzyme does not cleave a sequence because the 4 successive nucleotides are not present (or not at the right place, too short or too long T-RF), it might work for another one. So, that’s way you can increase the resolution of your T-RFLP, by potentially cleaving more sequences. Furthermore, some enzymes are not “good” for the amplicons you want to cleave, so you have to choose properly your enzyme(s) (main factor affecting T-RFLP results). The problem with using several restriction enzyme, is a bit similar than for dual labelling of primer (but not as bad) is that you can end up with 2 different T-RF for a specific sequence, because a specific sequence can have several restriction sites. So you can analyse your data separately and see if you get the same pattern/resolution but you could also merge your results from both enzymes (but only one primer) and do one analysis and see if you improve your data.       

To analyse your data, you analyse both primers separately. Then, you should probably try to analyse your data for one primer, either including both enzymes or for each enzyme. When you want to merge results from your 2 enzymes, what you can do is simply to copy and paste the data (after alignement) from one enzyme under the result from the other one. You could also merge before alignement but I will not go into details now why doing one or the other. I know it is extra work to do the analysis on the different combination, but you need to test if having 2 labelled primers, and multiple enzymes is useful or not and if you should keep going into that direction or if with 1 labelled primer and 1 enzyme, you already get the answer you want. I would expect that you do not need to label both primer, but using several enzymes is probably useful for AM fungi and you could either work on both enzymes separately or merge the results.  

All these information are often not properly stated in articles using T-RFLP, and it is confusing and ender the replicability of science…

Hope my reply will help you, and sorry for its length…