Article Text


Applications and advances of CRISPR-Cas9 in cancer immunotherapy
  1. An-Liang Xia1,2,
  2. Qi-Feng He1,2,
  3. Jin-Cheng Wang1,2,
  4. Jing Zhu3,
  5. Ye-Qin Sha3,
  6. Beicheng Sun2,
  7. Xiao-Jie Lu1
  1. 1 Department of General Surgery, Liver Transplantation Center, The First Affiliated Hospital of Nanjing Medical University, Nanjing, China
  2. 2 Department of Hepatobiliary Surgery, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, China
  3. 3 Nanjing Medical University, Nanjing, China
  1. Correspondence to Dr Xiao-Jie Lu, Department of General Surgery, Liver Transplantation Center, The First Affiliated Hospital of Nanjing Medical University, Nanjing 210009, China; 189{at}


Immunotherapy has emerged as one of the most promising therapeutic strategies in cancer. The clustered regularly interspaced short palindromic repeat (CRISPR)-associated protein 9 (CRISPR-Cas9) system, as an RNA-guided genome editing technology, is triggering a revolutionary change in cancer immunotherapy. With its versatility and ease of use, CRISPR-Cas9 can be implemented to fuel the production of therapeutic immune cells, such as construction of chimeric antigen receptor T (CAR-T) cells and programmed cell death protein 1 knockout. Therefore, CRISPR-Cas9 technology holds great promise in cancer immunotherapy. In this review, we will introduce the origin, development and mechanism of CRISPR-Cas9. Also, we will focus on its various applications in cancer immunotherapy, especially CAR-T cell-based immunotherapy, and discuss the potential challenges it faces.

  • crispr-cas9
  • genome editing
  • car-t cells
  • cancer immunotherapy
  • gene therapy

Statistics from


Cancer is a complex disease that derives from a variety of genetic and epigenetic changes, and is a major threat to human life and public health.1 2 Conventional strategies such as surgery, radiotherapy and chemotherapy have made significant advances in the treatment of cancer. However, high recurrence rates and chemotherapy/radiotherapy resistance often occur in patients with cancer, leading to poor prognosis. Therefore, new therapeutic strategies are still needed for patients with cancer.

During the past few years, immunotherapy has emerged as one of the most promising therapeutic strategies in oncology. Currently, it has been revealed that immune checkpoint inhibitors can achieve remarkable efficacy against a wide range of solid and haematological cancers through reversing dysfunctional or exhausted T cells.3 4 In addition, genetically engineered T cells such as the chimeric antigen receptor T (CAR-T) cells are another form of cancer immunotherapy. Some clinical trials using modified T cells have achieved great success in acute and chronic leukaemia,5 6 showing higher response rates and persistent response. Given this, gene editing of patients’ immune cells can be achieved through the clustered regularly interspaced short palindromic repeat (CRISPR)-associated protein 9 (CRISPR-Cas9) system.

The CRISPR/Cas9 system is to date the most efficient and flexible genome editing system and has been widely applied in various cell types and organisms to site-specifically edit single or multiple target genes.7–9 Unlike zinc-finger nucleases (ZFNs) and transcription activator-like effector nucleases (TALENs),10 11 CRISPR-Cas9 technology exploits the single-guide RNA (sgRNA) for site recognition, depending on the simple and accurate Watson-Crick base pairing between sgRNA and target DNA.12 Also, with its powerful gene editing efficiency and ease of use, it can be implemented to fuel the production of therapeutic immune cells. In this review, we will introduce the historical development and mechanism of CRISPR-Cas9. Then we will discuss its application in cancer immunotherapy particularly in CAR-T cell therapy, and the potential challenges it faces in the future.

Origin, development and mechanism of CRISPR-Cas9 system

The CRISPR-Cas9 genome editing system originates from the naturally occurring adaptive immunity found in a variety of bacteria (figure 1A). The study on CRISPR dates back to 1987, when Nakata and his colleagues13 first reported a surprising set of interspaced short repetitive sequences downstream of the Escherichia coli iap gene. However, the significance of these repetitive sequences was not very clear at that time. Over the next 15 years, increasing such repeats were also found in other bacteria and archaea, which aroused researchers’ various speculation about their possible functions. Until 2002, the term CRISPR was used to characterise this family of interspaced repeats for the first time.14 The CRISPR-associated (Cas) genes were identified and always located near the CRISPR locus, indicating that the Cas genes have potential functional relationship with the CRISPR loci.14 Barrangou et al 15 demonstrated experimentally the function of a type II CRISPR-Cas system in 2007, as an adaptive defensive system against phage infection. In 2013, two research teams presented groundbreaking studies, revealing the first application of the CRISPR-mediated genome editing in mammalian cells.16 17 This represented a great leap forward in genome editing. Nowadays, this technology has been used extensively to edit various eucaryotic genomes such as zebrafish, mice and monkeys.18–21

Figure 1

The origin, development and mechanism of CRISPR-Cas9 system. (A) Timeline of the origin and development of CRISPR-Cas9 system. (B) The mechanism of CRISPR/Cas9-mediated genome editing. The Cas9 nuclease is directed to the specific genomic sequences by sgRNA and then induces DSBs adjacent to the PAM. Subsequently, DNA repair machinery is initiated to repair DSBs through two pathways: NHEJ or HDR. NHEJ, as a primary and efficient repair mechanism, repairs lesions by connecting the two ends of the DSBs, and frequently gives rise to small insertions or deletions (indels) for gene knockout experiments. In contrast, HDR is an inefficient but more accurate repair mechanism because it requires the existence of a DNA template. This repair mechanism, based on the DNA template and position of homology arms, can be used for precise genome modification at the site of the DSBs for gene knock-in experiments. CRISPR-Cas9, clustered regularly interspaced short palindromic repeat-associated protein 9; DSBs, double-strand breaks; HDR, homology-directed repair; NHEJ, non-homologous end joining; PAM, protospacer adjacent motif; sgRNA, single-guide RNA.

The CRISPR-Cas9 system comprises two essential components: sgRNA and DNA endonuclease Cas9, with the former guiding the latter to target sequence to cut double-stranded DNA site-specifically (figure 1B). The Cas9/sgRNA complex seeks the complementary loci of the genome according to the principle of the Watson-Crick base pairing between the guiding sequence and the DNA. Once the target is successfully identified, the conformational changes will occur in Cas9, forming two nuclease domains.22 23 Then, the nuclease domains cleave both strands of the target DNA, located at approximately −3 nucleotides in front of protospacer adjacent motif (PAM), resulting in the generation of double-strand breaks (DSBs).24 The most common PAM of Streptococcus pyogenes Cas9 is ‘NGG’, with the advantage of frequent occurrence in the genome and minimal restrictions on target site selection and sgRNA design.25 Interestingly, single amino acid mutations in one of the two nuclease domains can bring about a ‘nickase’ that induces single-stranded DNA nicks rather than DSBs.26 27 Similarly, the two nuclease domains can mutate simultaneously, causing Cas9 to lose the ability to cut DNA, named catalytically inactive or ‘dead’ Cas9 (dCas9). The dCas9 can be modified with fusion proteins to activate or inhibit the transcription,27–30 or alter the epigenetic state of the local chromatin.31–34

When the DSBs are generated, the host’s DNA repair machinery is immediately activated to repair DSBs through one of two endogenous DNA repair pathways: non-homologous end joining (NHEJ) or homology-directed repair (HDR) (figure 1B).35 NHEJ, as a primary and efficient repair mechanism, repairs lesions by connecting the two ends of the DSBs, and frequently gives rise to small insertions or deletions (indels) at the breaking site without the existence of a DNA template. Target genes can be disrupted via these indels-mediated reading frame changes, causing mRNA degradation or non-functional protein production.36 Also, NHEJ can be used to restore the reading framework of dysfunctional genes to treat diseases such as Duchenne muscular dystrophy.37 38 In contrast, HDR is an inefficient but more accurate repair mechanism because it requires the existence of a DNA template. This repair mechanism, based on the DNA template and position of homology arms, can be used to correct dysfunctional genes and restore gene function.

Different genome editing platforms

Successful gene editing has the following conditions: a nuclease generating a DSB and complete DNA repair mechanisms. Most importantly, DSBs are generated at the desired target sequence by nucleases to avoid off-target effects. To date, four well-known platforms have been applied to generate site-specific DSBs in the genome of interest: meganucleases, ZFNs, TALENs and CRISPR-Cas9 system, with each having their respective advantages and limitations (figure 2).

Figure 2

Four different genome editing platforms. Meganucleases, ZFNs and TALENs achieve site recognition through protein–DNA interactions, while CRISPR-Cas9 enables site-specific recognition via RNA–DNA interactions. ZFNs and TALENs contain DNA-binding protein domains that can bind 3 or 1 nucleotide, respectively, in a sequence-specific manner. Also, different genome editing platforms have their own advantages and limitations. CRISPR-Cas9, clustered regularly interspaced short palindromic repeat-associated protein 9; PAM, protospacer adjacent motif; sgRNA, single-guide RNA; TALENs, transcription activator-like effector nucleases; ZFNs, zinc-finger nucleases.

Meganucleases are a class of endonucleases with extended DNA-binding sequences, relying on the capability of homing endonucleases to recognise and cleave DNA.39 For example, naturally occurring homing endonucleases include I-Crel and I-Scel enzymes, which can be engineered to extend DNA recognition sites.40 41 In addition, the recognition site of meganucleases is approximately 14 bp long with the increased specificity, and the size of the meganuclease is relatively small with the potential of in vivo delivery.39 42 However, the complexity of re-engineering these endonucleases and the lower editing efficiency have greatly limited the application of this platform, thus hindering the advancement of this technology.

ZFNs are artificial proteins with zinc finger DNA-binding domains (ZFDBDs) fused to a non-specific nuclease domain like FokI. ZFNs, as the first platform for successful human somatic genome editing, have also been used to create mutant animals.43–45 With rational design and use of combinatorial libraries, ZFDBDs can be further reprogrammed to target diverse DNA sites.46 ZFDBD recognises a triplet nucleotide code, and FokI endonuclease functions as a dimeric form to generate DSB in the DNA, all of which increase the sequence specificity of ZFNs.11 Nevertheless, ZFNs have gradually faded out because of the technical challenges related to the linkage engineering of protein domains and the need for two synthetic ZFDBDs with proper spacing and orientation.47

Similarly, TALENs consist of a specific DNA recognition domain fused to a non-specific FokI nuclease domain.48 The TALE protein, first discovered in the plant pathogen Xanthomonas, could specifically recognise and bind DNA.49 50 Despite the only restriction of a T presence at the 5’ end, TALENs can almost target any DNA sequence of interest.51 In addition, TALENs are very specific and not too cumbersome. However, because about 34 amino acids are needed to bind a single nucleotide, the TALENs exhibit a large size, thus hindering the in vivo delivery.

Finally, the CRISPR/Cas9 platform, as described above, has been recognised as the preferred genome editing technology without the need to reprogramme the protein domain for targeting new sequence.12 Through simply changing the guide RNA sequence, the Cas9 protein can retarget new sequence. Of course, by expressing different sgRNAs, the CRISPR system can simultaneously generate multiple DSBs.

Applications of CRISPR-Cas9 in cancer immunotherapy

Apart from being a formidable research tool, CRISPR/Cas9-mediated genome editing holds immense promise in cancer immunotherapy. Currently, immunotherapy is an extremely promising therapeutic strategy, especially in cancer. For instance, adoptive T cell therapy in cancer immunotherapy has demonstrated significant efficacy in treating cancer.52–54 This approach involves a series of processes in which the patient-derived immune cells are genetically modified in vitro and then reinfused into the patient to mediate specific recognition and killing of the tumour cells. As such, CRISPR-Cas9 can be applied to promote the mounting emergence of therapeutic immune cells due to its versatility and ease of use.

The generation of CAR-T cells is one of the most eye-catching applications of CRISPR-Cas9 technology in cancer immunotherapy. In general, CAR contains an intracellular chimeric signalling domain capable of activating T cells and an extracellular single-chain variable fragment that can specifically recognise tumour antigens.55 56 These genetically modified T cells carrying tumour-targeting receptors have achieved positive therapeutic outcomes in patients with various haematological malignancies such as leukaemia and lymphomas.57–59 Until now, CAR-T cell therapy targeting CD19 is the most effective due to its specific expression only in B cells and B cell malignancies.60–62 Currently, patients’ own T cells are the predominant source of producing CAR-T cells, called autologous CAR-T cells. The manufacture of autologous CAR-T cells is time-consuming and costly, involving the isolation, modification and expansion of individualised T cells. In addition, for some special populations including neonates, elderly or patients with cachexia, it is hard to acquire T cells with good quality and enough quantity to produce patient-specific CAR-T cells. These reasons have greatly hindered the widespread clinical applications of CAR-T cell therapy. Therefore, allogeneic universal T cells derived from healthy donors may be potential substitutes. Nonetheless, the endogenous TCR and HLA on allogeneic T cells need to be eliminated to avoid graft-versus-host disease and the rapid rejection of the host immune system. Given this, Ren et al 63 used the CRISPR/Cas9 system to simultaneously disrupt multiple genomic sites for constructing universal CAR-T cells with defective TCR and HLA class I expression, which shows potent antitumour activity. The Fas receptor, also known as CD95, can induce apoptosis of T cells and damage the CAR-T function once it is combined with its ligand FasL.64 Therefore, through the CRISPR-Cas9 technology, CAR-T cells with Fas knockout can better exert the tumour-killing ability in tumour-bearing mice and prolong their life span.65 Of note, one study showed that the use of CRISPR-Cas9 to introduce CAR into the T cell receptor α constant (TRAC) locus contributed to more uniform CAR expression in T cells and increased T cell potency with decreased terminal differentiation and exhaustion, thus immensely outperforming conventionally engineered CAR-T cells in a mouse model with acute lymphoblastic leukaemia.66

Aside from generating universal CAR-T cells, CRISPR-Cas9-mediated genome editing can also be applied to eliminate genes that encode inhibitory T cell surface receptors, such as programmed cell death protein 1 (PD-1) and cytotoxic T lymphocyte-associated protein 4 (CTLA-4),65 67 68 for improving efficiency of T cell-based immunotherapy in cancer. With Cas9-sgRNA ribonucleoprotein and exogenous single-stranded DNA template, targeted nucleotide substitutions have been attained at the PD-1 locus of primary T cells, resulting in enhanced T cell effector function.69 Similarly, a study by Su et al 68 indicated that PD-1 expression was significantly reduced via electroporation of sgRNA and Cas9 encoded plasmids into human T cell-mediated PD-1 gene disruption, which increased the killing ability of T cells against cancer cells. Indeed, the first clinical trial of CRISPR/Cas9 has been started, which uses CRISPR/Cas9 to mediate PD-1 knockout of T cells in patients with lung cancer, and CAR will not be introduced into T cells.70 Subsequently, similar clinical trials with PD-1-knockout autologous T cells are also under way for prostate cancer (NCT02867345), bladder cancer (NCT02863913) and renal cell carcinoma (NCT02867332). The safety and efficacy of PD-1 knockout modified T cells for cancer treatment will be evaluated. Moreover, Zhang et al 71 successfully generated lymphocyte activating gene-3 (LAG-3) knockout CAR-T cells using CRISPR-Cas9, confirming that no significant viability or immunophenotypic changes were observed in cultured CAR-T cells. This study showed that CAR-T cells with LAG-3 knockout possessed strong antigen-specific antitumour activity in a xenograft mouse model.71 Excitingly, with the one-shot CRISPR system, simultaneous gene editing of four gene loci to generate allogeneic universal T cells deficient of PD-1 and CTLA-4 was also accomplished.65

Furthermore, the CRISPR-Cas9 system, as a formidable genome editing tool, has been exploited to be a function-based, large-scale screening strategy in mammalian cells. Through this strategy, Manguso et al 72 successfully identified protein tyrosine phosphatase non-receptor type 2 (PTPN2) as a novel cancer immunotherapy target in mouse transplantable tumour models.

Challenges for completely therapeutic implementation

Although CRISPR/Cas9 technology has shown immense potential in cancer immunotherapy, it still faces various challenges for its complete transformation into clinical treatment. This requires the identification of the critical factors affecting the therapeutic results by CRISPR/Cas9-mediated genome editing and the development of effective strategies accordingly.

The first challenge should be to choose the optimal nuclease platform and rationally design the sgRNA, depending on the research and treatment requirements. As described above, four different genome editing platforms provide corresponding advantages and limitations, in terms of in vivo delivery, site specificity and flexibility. The process of meganucleases and ZFNs targeting new sequences is time-consuming, requiring 1 year or 3 months to engineer complex proteins, but TALENs require about 2 weeks for engineering. Compared with the previous, CRISPR-Cas9 can retarget novel DNA sequences through simple sgRNA sequence alterations. Also, CRISPR-Cas9 can introduce multiple DSBs simultaneously by expressing different sgRNAs, thereby enabling more intricate gene editing.16

The second challenge is to safely and effectively deliver gene editing tools into target cells in vivo or in vitro. Multiple elements must be transported into one cell nucleus. For ZFNs and TALENs, it needs a pair of proteins. As for CRISPR-Cas9, it needs the Cas9 protein and guide RNA, and for precise gene editing, an additional donor DNA template may be also essential. In general, both viral and non-viral vectors can be exploited to deliver Cas9 as well as sgRNA. Delivering one or a limited number of programmable nucleases in a transient manner may provide better clinical results.

The third challenge is to ensure efficient genome editing after editing components into the nucleus. Improvements in editing efficiency with reduced off-target effects will be imperative for achieving better overall therapeutic efficacy. In different cell types and different cell states, the efficiency of NHEJ-mediated and HDR-mediated DSB repair distinctly varies. NHEJ usually generates indels at the cleavage site, while HDR, under the premise of DNA template, replaces the targeted allele via recombination with the alternative sequence. HDR templates are typically single-stranded oligonucleotides or plasmids containing alleles adjacent to target sites, which also require to be transported into the nucleus. And viral or non-viral vectors have been successfully applied to HDR template delivery.73 In theory, nanocarrier-based HDR template delivery involves the following steps: formulation by electrostatic interaction, maximisation of cellular uptake, prevention of degradation, and translocation into the cellular nucleus, which resembles the delivery of other oligonucleotides or plasmid DNA. Of course, these criteria are analogous to Cas9 plasmid/protein and sgRNAs delivery. Therefore, to increase HDR efficiency, we may combine the nanocarriers used for HDR templates with Cas9 plasmid/protein or sgRNAs to promote the codelivery of Cas9 components/sgRNAs and HDR templates.

The fourth challenge is to reduce the potential off-target effects with increased CRISPR-Cas9 specificity. So far, two general approaches have been introduced to reduce the potential off-target effects of engineered nucleases. One is to increase the specificity of nuclease-mediated cleavage of the target site, and the other is to restrict the duration of nuclease expression to reduce the possibility of off-target mutations. For example, compared with plasmids and viruses, efficient gene editing with Cas9 mRNA and protein reduces off-target effects, probably because mRNA or ribonucleoproteins degrade rapidly after targeted cleavage.74 In addition, molecular research may provide novel mechanistic insights for target recognition and cleavage of Cas9-gRNA complexes, which can provide more possibilities for improving the specificity.

The fifth challenge is the fitness of edited cells and the therapeutic threshold of editing.75 If the edited cells possess a better capability to proliferate or show greater adaptability than the unedited cells, this will facilitate the edited product to reach the therapeutic threshold required for successful treatment outcomes. In contrast, under the condition of low editing efficiency or lack of adaptability of edited cells compared with unedited cells, the therapeutic outcome is not what we expected. In view of this, the obstacle may be partially overcome through in vitro genome editing, and the edited cells can be reinfused into the patients after being expanded to the appropriate amount as much as possible.

The last potential challenge is the immune response evoked by Cas9 nuclease, a bacterial-derived foreign protein that has been validated.76 77

Concluding remarks

In summary, genome editing with CRISPR-Cas9 holds immense therapeutic potential for improving T cell-based immunotherapy. T cell therapy based on genome editing, however, still faces some challenges for complete clinical applications. The issue of priority is the safety of genetically engineered T cells. In view of this, some studies have attempted to increase the specificity of gene editing with minimised off-target effects, but the degree of accuracy still needs to be determined for specific clinical applications. In addition, it is not clear how the autoimmune system will be in response to genetically engineered cells. Excitingly, extensive research and encouraging results from the use of universal CAR-T therapy in clinical applications have shown that T cell therapy with CRISPR-Cas9 technology holds great promise in cancer treatment. Therefore, although with some challenges, we can envision that the continued advancement of CRISPR-Cas9 technology will contribute to its therapeutic application in cancer immunotherapy in the foreseeable future.


View Abstract


  • A-LX, Q-FH and J-CW contributed equally.

  • Contributors XJL conceived the idea. A-LX drafted the manuscript. Q-FH, J-CW, JZ and Y-QS contributed to performing the literature collection. BS and X-JL directed and approved the manuscript. All the authors gave the final approval of the manuscript submission.

  • Funding This work was supported by grants from the National Natural Science Foundation (grant number: 81772596 to X-JL).

  • Competing interests None declared.

  • Patient consent Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.