Tianli Hui,

All the answers above are great points to consider.  

Ultimately, you are looking for consistency and reproducibiility in your results.  Ask yourself can someone repeat this experiment and get the same result?  Have I given enough experimental detail for the experiments to be repeated?  In practice and in addition to the points above, this is how I view things:  

1) Transient transfection/knock down effects are acute - they give clues to function in cell processes but are not suitable for disease modeling where typically a cell has adapted to mutation, gain or loss of function of a gene.  Transient experiments are perfectly suitable for determining initial cell processes a gene product is in involved in and looking for changes in protein interactions or gene expression under acute conditions of knock-down or over-expression.   Also, consider that transient and stable short term (24-72 h) knock-down using viral transduction of your cells may give similar results and by keeping your pool of cells in culture longer (2-8 weeks) you can also observe adaptation to the knock-down of your gene. 

2) Stable transfection is most suitable for disease process modeling with the caveat that if you are making single clonal cell lines you have to make several to account for normal genetic drift and/or genetic mosaicism in your cell culture or tumour.  In this case, you may want to make a stable population of cells where you do not select single clones.  Lastly, stable knock-down or over-expression phenotypes also drift over time, so you want to do all your experiments at the same passage number range to avoid genetic adaptations or drift that will affect reproducibility.  For some genes this can happen even after freezing down cells, so you may have to derive all data immediately from your population of cells and repeat the experiment for replicates, rather than freezing aliquots of cells and repeating later on from thawed samples.  If  you are using shRNA and not CRISPR knock outs, you may also see silencing of your shRNA in stable cell lines if your shRNA expression reduces cell fitness or affects the cell cycle. In this case you can only observe short term transient effects of knock-down, and the adaptation is not "REAL" adaptation but an artifact of silencing of the shRNA.

3) Be aware of off-target or "tag" specific effects of knock-down and over-expression (respectively) in your experiments.  Ideally you want to use multiple approaches, such as shRNA with at least two shRNAs, and confirm by add-back of the cDNA and/or CRISPR knock-out if you are studying a non-essential gene.  For over-expression you want to use  no-tag if possible,  or if you are using a fluorescent protein, repeat experiments with no tag, or with HA or Flag (or similar small tag).  If you are tagging your protein for the first time, and there is no information in the literature, tag both ends of your protein in separate experiments and make sure you have an assay for function or proper localization of your protein.  Also be aware of the level of expression in add back experiments, you don't want more than 2 fold over endogenous expression before knock-down.  

4) Be aware of what isoform of your protein you are using for add back  experiments or if your shRNA is only targeting a single isoform of your gene transcript.  This is a HUGE source of variability between labs and studies that is not always accounted for. IF there are multiple isoforms of your gene you will have to test phenotypes with add back of at least the two predominant isoforms based on their expression in your cell model.  With so much information available now on splice isoforms of every gene, I don't think you can plead ignorance anymore on this issue and more and more papers will be asked to be revised or will be rejected if it is not clear which isoform you are using in your experiments. 

In the end, reproducibility over time and in other cell lines is the key guiding principle you want to follow to produce the best data.  In my view as long as your results are reproducible, you have used more than one assay and cell line, and you are clear about how you obtained your data, the results you produce are valid and will stand the test of time.

Best of luck in your experiments.

Doc Dellaire

Depends on your experiment and cell type. If efficiency of  transfection is high, then for most of your purpose transient experiment is enough. If your cells don't transfect at high rate then i suggest you to do stable transfection. The advantages of transient transfection is the speed of the experiment. The disadvantage of stable  transfection is the time to for selection and characterization of clones before you conduct the experiment of your interest

best approach is transient transfection because you can get a quick answer and it's also a way to test the capabilities in your lab. If you can't get a transient transfection to work, committing to stable transfection would be unacceptable.

Thanks Julio and Oscar for this answer. in my mind , when you get the stable transfected cell, trarget gene function will not be acheived as  the cell has adpted the change. is that right?

usually there is  redundancy of gene functions that allows adaptation or survival of the cells if your target gene is not essential. But there should be changes that you could detect if you has an idea of the function. The changes  will happen in both transient or stable cells

@Julio, it means both transient and stable cells should be tested for well? thank you!

No, if you get 90% of transfection efficiency then the transient transfection is ok ns Oscar mention you get the results quickly. If the efficiency is low then you should try the stable transfection

Tianli Hui,

All the answers above are great points to consider.  

Ultimately, you are looking for consistency and reproducibiility in your results.  Ask yourself can someone repeat this experiment and get the same result?  Have I given enough experimental detail for the experiments to be repeated?  In practice and in addition to the points above, this is how I view things:  

1) Transient transfection/knock down effects are acute - they give clues to function in cell processes but are not suitable for disease modeling where typically a cell has adapted to mutation, gain or loss of function of a gene.  Transient experiments are perfectly suitable for determining initial cell processes a gene product is in involved in and looking for changes in protein interactions or gene expression under acute conditions of knock-down or over-expression.   Also, consider that transient and stable short term (24-72 h) knock-down using viral transduction of your cells may give similar results and by keeping your pool of cells in culture longer (2-8 weeks) you can also observe adaptation to the knock-down of your gene. 

2) Stable transfection is most suitable for disease process modeling with the caveat that if you are making single clonal cell lines you have to make several to account for normal genetic drift and/or genetic mosaicism in your cell culture or tumour.  In this case, you may want to make a stable population of cells where you do not select single clones.  Lastly, stable knock-down or over-expression phenotypes also drift over time, so you want to do all your experiments at the same passage number range to avoid genetic adaptations or drift that will affect reproducibility.  For some genes this can happen even after freezing down cells, so you may have to derive all data immediately from your population of cells and repeat the experiment for replicates, rather than freezing aliquots of cells and repeating later on from thawed samples.  If  you are using shRNA and not CRISPR knock outs, you may also see silencing of your shRNA in stable cell lines if your shRNA expression reduces cell fitness or affects the cell cycle. In this case you can only observe short term transient effects of knock-down, and the adaptation is not "REAL" adaptation but an artifact of silencing of the shRNA.

3) Be aware of off-target or "tag" specific effects of knock-down and over-expression (respectively) in your experiments.  Ideally you want to use multiple approaches, such as shRNA with at least two shRNAs, and confirm by add-back of the cDNA and/or CRISPR knock-out if you are studying a non-essential gene.  For over-expression you want to use  no-tag if possible,  or if you are using a fluorescent protein, repeat experiments with no tag, or with HA or Flag (or similar small tag).  If you are tagging your protein for the first time, and there is no information in the literature, tag both ends of your protein in separate experiments and make sure you have an assay for function or proper localization of your protein.  Also be aware of the level of expression in add back experiments, you don't want more than 2 fold over endogenous expression before knock-down.  

4) Be aware of what isoform of your protein you are using for add back  experiments or if your shRNA is only targeting a single isoform of your gene transcript.  This is a HUGE source of variability between labs and studies that is not always accounted for. IF there are multiple isoforms of your gene you will have to test phenotypes with add back of at least the two predominant isoforms based on their expression in your cell model.  With so much information available now on splice isoforms of every gene, I don't think you can plead ignorance anymore on this issue and more and more papers will be asked to be revised or will be rejected if it is not clear which isoform you are using in your experiments. 

In the end, reproducibility over time and in other cell lines is the key guiding principle you want to follow to produce the best data.  In my view as long as your results are reproducible, you have used more than one assay and cell line, and you are clear about how you obtained your data, the results you produce are valid and will stand the test of time.

Best of luck in your experiments.

Doc Dellaire