In the realm of single-cell sequencing, one of the biggest messages that must not be ignored is ‘garbage in, garbage out.’ In more technical terms, taking steps to guarantee the highest quality of cellular input can significantly improve the quality and validity of results. The initial quality of a sample is directly related to variables in the storage, dissociation, and purification processes. One of the most important considerations, particularly for tissue samples, is storage parameters.
While many protocols have undergone optimization for sample storage, the gold standard remains fresh samples. The need for alternatives is largely driven by practical concerns in the context of clinical samples. For instance, when processing patient tissue samples that often arrive in the evening following surgeries, traditional techniques for single-cell profiling would demand hours of continuous work before reaching a safe stopping point. Accordingly, researchers have resorted to freezing or fixing samples, which introduces certain limitations and may alter data.
It is important to note that single-nucleus RNA-seq (snRNA-seq) is more compatible with freezing than single-cell RNA-sequencing (scRNA-seq), but many researchers have a preference for exploring cytoplasmic transcripts to glean maximal cellular information. Additionally, single-cell sequencing – as opposed to snRNA-seq – is the technique of choice for specific cell types, particularly immune cells.
The impact of fixation or freezing on single-cell data
Transcripts are labile molecules that are highly responsive to environmental factors and stressors. Accordingly, sharp shifts in environmental conditions can lead to major shifts in the transcriptome, thereby yielding results that are no longer representative of the cell’s natural state. This is also true for cells that have lost extensive material due to RNA degradation and other potential damage. In recent years, as researchers have attempted to develop techniques for sample preservation prior to single-cell experiments, many significant differences have emerged in downstream results.
One of the reported differences in fresh versus frozen or fixed samples is a decrease in the number of genes and/or cells in the latter. For instance, in an scRNA-seq protocol in which retinal cells were fixed with methanol, the median number of genes detected per cell was lowered by 500 to 1,000 genes when compared to the number detected in cells derived from fresh samples.[1] Similarly, the cDNA yield was lower in cells fixed with dithio-bis(succinimidyl propionate) (DSP) – also known as Lomant’s Reagent – than in fresh cells.[2]
The number of cells recovered can also vary widely after freezing and this can significantly alter conclusions generated based on cell clustering during data analysis, particularly since specific cell types may be selectively depleted. In one study, CD45+ cells, mast cells, and granulocytes were present in lower amounts after freezing compared to fresh samples, regardless of freezing method.[3]
Studies have also consistently revealed a bias in transcript coverage towards the 3’ end. This has been found to increase with storage time.[4] In addition, longer transcripts and transcripts with higher GC content are significantly less abundant in fixed samples (methanol) than fresh samples.[5]
All cell populations are not affected by the single-cell preparation equally
Cells are known to be sensitive to changes in temperature and one major consequence of cryopreservation is therefore cell damage. Since the extent of sensitivity to temperature varies across cell types, this can impact the recovery of specific clusters. To characterize differences across cells, investigators compared the proportion of damaged cells that were fresh or cryopreserved at -80°C or in liquid nitrogen. In the fresh condition, only 14%, 2%, and 15% of HEK293, K562, and NIH3T3 cells were damaged, respectively. In contrast, upon freezing at -80°C, the proportion of damaged cells was 66%, 55%, and 20%, respectively. Similarly, freezing in liquid nitrogen led to cell damage rates of 61%, 49%, and 17%, respectively. These trends were also confirmed in cells from mouse colon samples.[6]
Importantly, some studies have pointed towards a change in gene expression in the final stages of analysis. In a study on scRNA-seq on fresh or frozen microglia, there were 538 differentially expressed genes (DEGs) between the two conditions.[7] Many of these genes were found to be involved in the acute phase response and included genes that reflect inflammatory status, such as IL6 and TNFR2, which are relevant outputs for microglial function and reactivity. Here, the changes in these critical genes as a result of processing can be mistaken for biological changes. Immediate-early genes also exhibited more than a 2-fold increase in frozen cells compared to fresh cells.
Plan ahead of your single-cell experiment
The disadvantages associated with using frozen or fixed samples in single-cell RNA sequencing experiments must not be overlooked. The introduction of artifacts related to storage parameters, rather than real characteristics of biological samples, creates major concerns regarding the accuracy of the resulting data. Moreover, extensive loss or damage of cellular material during preservation and handling can yield a transcriptome that is no longer representative of that of the original cell. Changes in gene expression levels are particularly problematic in the context of comparative analyses between control and treatment conditions as the results may reflect technical parameters rather than treatment-related effects.
While researchers have had to settle for a tradeoff between practicality and quality due to lengthy processing workflows, recent techniques that offer safe stopping points early in processing preclude the need to compromise the integrity of the samples. To maximize the accuracy and precision of scRNA-seq experiments and ensure the relevance and reliability of findings, fresh samples should be used whenever possible.
Bibliography
[1] https://www.ncbi.nlm.nih.gov/p...
[2] https://www.ncbi.nlm.nih.gov/p...
[3] https://www.sciencedirect.com/...
[4] https://www.ncbi.nlm.nih.gov/p...
[5] https://bmcgenomics.biomedcent...