Reduce mitochondrial reads in scRNA-seqwithout losing a critical QC signal.

Elevated mitochondrial fractions, defined as the proportion of sequencing reads or UMIs mapping to mitochondrial encoded genes, are commonly observed in single-cell RNA sequencing (scRNA-seq) datasets, reflecting both underlying biological states and sample handling effects^1-3. From an analytical perspective, mitochondrial reads contribute little to the resolution of most scRNA-seq studies. They are typically excluded from highly variable gene selection and are rarely used in differential expression analyses. This treatment is explicitly reflected in widely used workflows such as the Seurat PBMC tutorial and Scanpy’s preprocessing guidelines^4,5.

Sequencing capacity allocated to mitochondrial transcripts reduces the effective depth available for genes that define cell identity, state, and function. At the same time, the proportion of mitochondrial transcripts is widely used as a quality-control metric, where high levels are interpreted as an indicator of cellular stress, damage, or loss of cytoplasmic RNA⁶. This rationale is discussed in detail in the Bioconductor’s Orchestrating Single-Cell Analysis tutorial⁷.

Any strategy aimed at modifying mitochondrial read abundance must therefore provide gains in information content while preserving quality control interpretability. In this blog we introduce a CRISPR/Cas9-based approach to deplete mitochondrial-derived reads at the dsDNA stage of any workflow, a new approach that has emerged in response to these requirements.

Why mitochondrial reads remain a limiting factor in scRNA-seq datasets

15-30% of reads mapping to mitochondrial transcript is frequently observed in datasets derived from primary tissues, dissociated solid tumors, activated immune populations, and other clinically heterogeneous samples1-3. The resulting loss of usable capacity impacts gene detection rates weakens sensitivity for low- and intermediate-abundance transcripts and can obscure transcriptional programs that distinguish closely related cell states. In discovery-oriented studies or analyses focused on rare populations, these effects can influence biological conclusions. An additional consideration is the contribution of nuclear mitochondrial pseudogenes. These mitochondrial-derived sequences embedded in the nuclear genome can create ambiguous alignments in short-read scRNA-seq data and inflate apparent mitochondrial signal or depending on alignment and referent, further complicating interpretation of mitochondrial fractions⁸.

Cas9–gRNA–based strategies such as the NEXTFLEX™ Cas9-gRNA Mito Depletion Enzyme have been developed to address the disproportionate contribution of transcripts originated from mitochondrially encoded genes and nuclear mitochondrial pseudogenes in scRNA-seq libraries. Rather than discarding reads computationally after sequencing, this approach removes targeted fragments at the library level, redistributing sequencing capacity toward transcripts that contribute directly to downstream analysis without altering upstream sample handling. In a recent case study with this method, depletion reduced mitochondrial read abundance by approximately 7- to 54-fold, depending on sample type.

The mitochondrial fraction is a flexible, context-dependent quality metric rather, and it is best interpreted relative to the distribution of the dataset rather than against a fixed threshold. While filtering is often applied in the ~15–25% range in droplet-based scRNA-seq, these values are not universal and are known to vary substantially with tissue type, cell state, dissociation method, and library preparation protocol. Adaptive approaches that jointly consider mitochondrial fraction and gene complexity, such as miQC, further underscore the importance of relative rather than absolute thresholds⁶.

Partial depletion shifts downward mitochondrial fraction while preserving relative differences between cells. Cells previously high in mitochondrial content often remain high relative to their peers, albeit on a different numerical scale. This means that the mitochondrial quality control can still be utilized, simply requiring recalibration of thresholds based on the depleted data distribution rather than applying historical cutoffs derived from undepleted libraries.

Applications where mitochondrial depletion would have the greatest impact

The benefits of mitochondrial depletion are most apparent in experimental settings where there are small transcriptional differences that are easily masked by limited sequencing depth. In these contexts, reallocating reads from mitochondrial transcripts to nuclear genes improves the interpretability and resolution of downstream analyses⁹.

One example is the analysis of early activation and stress-response programs in immune cells. Transient transcriptional states such as early T cell activation, interferon responses, or stress-induced signaling pathways are often characterized by modest changes in low- to intermediate-abundance nuclear transcripts rather than large shifts in dominant markers. In other words, immune activation is not dominated by a few highly expressed genes¹⁰. In these settings, metabolically active immune populations can have elevated mitochondrial signal11are,. Partial depletion would potentially increase sensitivity for these transcriptional programs, enabling detection without forcing aggressive mitochondrial filtering that would otherwise remove the very cells of interest.

Mitochondrial depletion would also be impactful in studies focused on rare or transitional cell populations, such as progenitor states, early differentiation intermediates, or small malignant subclones in tumor samples. These populations are typically underrepresented in the data and, in certain contexts, can exhibit elevated mitochondrial signal associated with metabolic reprogramming or sample handling stress.^3,12,13. By improving nuclear transcript coverage at fixed sequencing depth, depletion increases the likelihood that such populations can be resolved as distinct clusters and characterized transcriptionally, rather than being lost to noise or quality control filtering.

A further important application is the analysis of clinically derived and heterogeneous samples, including solid tumor biopsies, inflamed tissues, and cryopreserved material. These samples frequently exhibit broad distributions of mitochondrial fraction, reflecting variable cell viability, dissociation sensitivity, and metabolic state. Analysts are often forced into a trade-off between stringent mitochondrial thresholds, which preferentially remove tumor cells or stressed but viable populations, and permissive thresholds, which reduce overall analytical resolution. Recent work has demonstrated that rigid mitochondrial filtering can systematically bias cancer scRNA-seq studies by depleting metabolically altered but viable malignant cells, underscoring the limitations of fixed cutoffs^2,3.

Conclusion

These examples illustrate that the value of mitochondrial depletion extends beyond generic gains in sequencing efficiency. Its impact is expected to be greatest in precisely those applications where biological interpretation is most challenging. In these contexts, the use of Cas9-gRNA-based strategies such as NEXTFLEX™ Cas9-gRNA Mito Depletion Enzyme enhances resolution while preserving the flexibility needed for context-aware quality control.