For the generation of genetically engineered T cells, leukocytes are collected either from patients (autologous) or healthy donors (allogeneic). Notably, all FDA-approved CAR-T cell therapies to date are using patient-derived autologous cells but there is a number of clinical trials investigating the use of donor-derived allogeneic cells (2). Leukapheresis is the method of choice for the collection of leukocytes due to its availability at health care centers, patient tolerability, and its high yield of T cells for manufacturing (27, 28). Collected cells are either used fresh for direct manufacturing or more commonly cryopreserved for later handling. Cryopreservation takes place at the clinical center or in some cases at the manufacturing center. Although, cryopreservation has an impact on cell viability, on-site manufacturing of genetically engineered T cells is often not feasible – with the exception of a few academical clinical studies (NCT03676504 (9)) – and CAR-T cell generation can be achieved with frozen cells (29). There are differences of the cryopreservation procedure across clinical and manufacturing centers for different T cell products, regarding freezing media composition and durations, but their impact on the final product quality has not yet been comparably assessed.

The cellular composition of the starting material is paramount for the success to engineer a functional CAR-T cell product with long-lived antitumor properties. Enrichment of T cells from the leukapheresis product can be achieved with magnetic separation beads and all approved CAR-T cell therapies to date either use CD4⁺ and CD8⁺ T cells in a combined or a separate culture setting. Thus, CAR-T cell therapies are not limited to only generating CD8⁺ CAR-T cell responses but also utilize CD4⁺ CAR-T cells to improve their antitumor response in a synergistic fashion (30, 31). CD4⁺ T cells show more plasticity compared to CD8⁺ T cells and are comprised of T helper (Th) subsets and regulatory T (Treg) cells (32). Perturbation of CD4⁺ subsets and generation of Treg cells in particular might have an impact on the clinical response. In a recent study, the expansion of CAR-Treg cells has been associated with resistance to CD19 CAR-T cell therapy (33). Therefore, differentiation of CD4⁺ T cells to an “optimal” antitumor phenotype and limited generation of CAR-Treg cells during manufacturing might be essential to improve ACT. Separate manufacturing and administration of CD4⁺ and CD8⁺ T cells is already applied for the CD19 CAR-T cell therapy Breyanzi to reduce variability between the CD4⁺ and CD8⁺ CAR-T cell composition and to administer it in a dose-defined manner (34). In a clinical trial of a CD22 CAR-T cell therapy in children and young adults with CD22⁺ B-cell neoplasms, the change of the manufacturing protocol to CD4⁺ and CD8⁺ T cell selection improved manufacturing feasibility and reduced variability, however, this led to increased inflammatory toxicities and warranted dose de-escalation (35). The overall complete remission rate was still very high with 70% but may indicate CAR-T cell-induced toxicities based on the ratio of CD4⁺/CD8⁺ cells in the CAR-T cell product. In some cases separate manufacturing of CD4⁺ and CD8⁺ CAR-T cells could compromise CAR-T cell expansion. This has been the case in a third-generation CD20 CAR-T cell therapy where changing from CD4⁺/CD8⁺ selection to a combined culture setting improved manufacturing feasibility and clinical response rates (36). Moreover, enrichment and differentiation of specific memory T cell subsets, like multipotent T memory stem (TSCM) cells, may improve the antitumor responses of CAR-T cells (37, 38). This might be due to a certain level of stemness of the T cells that comes with higher T cell persistence and less susceptibility to exhaustion (39–41). Overall, the impact of the cellular composition during manufacturing on the antitumor efficacy of T cell therapies still needs to be compared in future clinical studies.

T cells are activated for efficient gene transfer and expansion which is commonly achieved by using anti-CD3/CD28 paramagnetic beads for viral transduction (42). However, this approach was reported to favor the expansion of CD4⁺ T cells over CD8⁺ T cells in non-enriched products (43) and could lead to even inefficient expansion of CD8⁺ T cells in some cases (44). This is likely due to the importance of CD28-mediated signaling in CD4⁺ T cells while 4-1BB costimulation is superior for the expansion of CD8⁺ memory T cells (44). To overcome this, current protocols have introduced cytokine cocktails in addition to anti-CD3/CD28 beads to support the expansion and to skew the differentiation into a phenotype with inherent good antitumor characteristics (45–49). For example, the cytokine IL-2 is used in standard protocols for its mitogenic effects on T cells and its potential benefits on T cell effectiveness in the context of tumor immunity (50).

In addition, the activation conditions and strength of the stimuli during manufacturing could also determine if the genetically engineered T cells are prone to exhaustion upon encountering their cognate antigen on the tumor cells. Soluble anti-CD3 antibodies together with mononuclear cells have been shown to result in a similar expansion efficacy of CD8⁺ T cells compared to anti-CD3/CD28 beads but induced a less terminally differentiated phenotype as well as less antigen-induced cell death and more expansion in previously activated CD8⁺ T cells (51). Acquisition of terminal differentiated effector functions during manufacturing actually may lead to impaired antitumor immunity in vivo (52). Due to the sensitive nature of T cell activation and differentiation and their impact on antitumor immunity and longevity of the genetically engineered T cells, new methods are constantly investigated. The Expamer technology is an interesting new approach for time-controlled initiation and termination by using soluble Strep-Tactin multimers that can be assembled with Twin-Strep-tag conjugated anti-CD3 and anti-CD28 Fab fragments and dissociated by adding non-toxic D-biotin (53). Soluble addition and inactivation of Expamer components for T cell activation is particularly attractive for large-scale production and might be useful to avoid overstimulation and subsequent apoptosis. On the other hand, less rigid surfaces for immobilization of anti-CD3/CD28 binders might be a better alternative since they have been shown to induce higher IL-2 production and expansion of CD4⁺ and CD8⁺ T cells ex vivo (54). Consistent with this, an antigen-presenting cell (APC)-mimetic scaffold that consists of a fluid lipid bilayer supported by mesoporous silica micro-rods and attached with anti-CD3/CD28 antibodies promoted two- to tenfold greater expansion compared to anti-CD3/CD28 paramagnetic beads (55). However, lipid bilayer systems might be less appealing compared to bead-based approaches for activation when it comes to large-scale manufacturing due to their increased technical complexity regarding handling and ease of removal. More studies that compare the effects of different activation conditions on the final T cell product for ACT will be necessary in the future in order to understand their impact on the clinical efficacy.

Viral vectors are used for all approved CAR-T cell therapies and most clinical trials for CAR- and TCR-T cell therapies due to their efficiency for stable gene transfer (56). Lentiviral (LV) and γ-retroviral vectors are the vectors of choice for cell engineering since they can carry larger genetic constructs and integrate the target gene into the genome of the engineered T cells compared to other viral vectors (57, 58). However, streamlining the generation of large quantities of viral vectors for manufacturing is a challenge and ensuring that no residual viral vectors or accidently transduced malignant cells are given to the patients comes with extensive and costly safety testing (59, 60). This presents difficulties for scaling out manufacturing and for making these therapies more affordable to meet the increasing demand for ACT. Therefore, non-viral approaches are currently investigated in early trials. Transposon-based gene delivery approaches, such as Sleeping Beauty (SB) or PiggyBac (PB) transposons, are cheaper and can carry larger genetic constructs compared to viral vectors while still integrating their target gene (61). SB transposition has already been used successfully for the manufacturing of CD19 or SLAMF7 CAR-T cells in early clinical trials without severe toxicity (62–65). In addition, automation of SB transposition was feasible and could be very attractive for large-scale manufacturing (25). Although PB transposition was also successful for the manufacturing of CAR-T cells and even showed an inclination to promote the generation of desired TSCM CAR-T cells (66, 67), a recent clinical trial observed the formation of CAR-T cell lymphoma in two out of ten patients (68, 69). This presented the first cases of malignant lymphoma derived from CAR-T cells and the investigators of the study caution for regular follow-ups of the patients receiving CAR-T cell therapies, especially when new methods for gene transfer are applied (69). The underlying causes for the observed malignant transformation in this study using the PB system is not fully understood yet. Insertional mutagenesis did not seem to be the cause since the pattern of integration of the CAR transgene was comparable to other studies using PB and in line with studies using viral systems (68). However, transcriptional upregulation of surrounding regions by the transgene promotor was observed but how these alterations may be involved in malignant transformation needs to be further addressed. The authors of the study do not think that this finding is an inherent problem with the PB system but rather it might be based on their manufacturing methodology with high-voltage electroporation and high concentration of transposon and transposase (69). Understanding the underlying mechanisms will help to develop safer manufacturing protocols and better safety readouts in the future.

Genome editing is a promising novel approach not only for the generation of genetically engineered autologous but also for “off-the-shelf” allogeneic T cells that might solve manufacturing challenges and excessive cost that are inherent to autologous T-cell therapies. In particular, clustered regularly interspaced short palindromic repeat (CRISPR)-Cas9 has already been used for the generation and clinical application of CAR-T cells with tolerable adverse events in cancer patients (70, 71). Multifactorial genome editing holds great potential for ACT with genetically engineered T cell in the future but will need further optimization and extensive safety monitoring to assess the risks for harmful off-target events. More information about CRISPR-Cas9 genome editing for the generation of engineered T cells is reviewed in (72, 73).

An important avenue for optimizing the manufacturing process of genetically engineered T cells is reducing the manufacturing time. This will reduce the cost and will scale-up manufacturing due to a faster turn-around of the engineered T cells. Most importantly, it might reduce mortality of patients with rapidly progressing cancers by reducing the vein-to-vein time. Standard protocols for CAR-T cell therapy culture CAR-T cells for 11 to 24 days which leads to a high number of harvested CAR-T cells (45–48). Interestingly, reducing the culture time of CD19 CAR-T cells to only 3 days increased their antitumor activity even at a 6-fold lower dose in a human xenograft model of acute lymphoblastic leukemia (ALL) (74). This could be due to an enriched proportion of stem-like T cells in the CAR-T product at reduced culture times. Remarkable manufacturing times haven been achieved with the FastT CAR-T next-day manufacturing platform, that was recently evaluated in a clinical trial for B-cell ALL (NCT03825718 (75)). Next day manufacturing with activation, LV transduction, and without expansion was feasible for all 25 patients with a tolerable safety profile and promising efficacy. Moreover, CD19 FasT CAR-T cells showed less exhaustion and a younger cellular phenotype compared to conventionally manufactured CAR-T cells in vitro but evaluation in larger clinical studies is needed. In addition, the T-Charge platform was used for manufacturing of CD19 CAR-T-cells in a phase I study with promising efficacy and safety profile (NCT03960840 (76)). The manufacturing time was less than 2 days and culturing time only took 24 hours. This approach also preserved naïve T and TSCM cells in the final CAR-T cell product which might increase the persistence of the genetically engineered T cells in patients. Rapid manufacturing of CAR-T cells was even demonstrated without activation and expansion within 24 hours and showed improved anti-leukemic activity in mouse xenograft models when compared to their conventionally manufactured counterparts (77). These new approaches seem to have great potential to reduce the vein-to-vein time which would greatly benefit the patients. However, since most of the T cell expansion takes place in the patients, this could mean that adverse events might be more difficult to predict in a temporal fashion upon therapy administration. Therefore, these new methods will have to be thoroughly tested in clinical trials with rigorous monitoring to ensure non-inferiority in efficacy and safety in comparison to standard long-time manufacturing procedures. Improving and standardizing manufacturing protocols for genetically engineered T cells, with every step potentially having an extensive impact on the antitumor activity and safety, is especially difficult since in-depth manufacturing protocols of approved CAR-T therapies are not publicly available. This will present a challenge for the future but overcoming this roadblock by greater exchange could fuel innovation and accessibility of these groundbreaking new therapies.

Immunotherapies with genetically engineered T cells are often accompanied by moderate to severe adverse events, limiting their clinical application in certain cases. The most common and severe adverse event of therapies with engineered T cells is cytokine release syndrome (CRS), which was first observed in clinical studies and did not occur in preclinical models at the time (113). Moreover, immune effector cell-associated neurotoxicity syndrome (ICANS), often referred as neurotoxicity, is very common in patients receiving CD19 CAR-T cells (114). Symptoms of CRS range from mild fever to life-threatening manifestations up to multi-organ system failure (115). Follow-up of patients receiving engineered T cells and monitoring for signs of CRS is vital to manage moderate to severe cases of CRS with anti-IL6 receptor agonist tocilizumab alone or in combination with corticosteroids, together with extensive supportive care (116). Symptoms of ICANS include headache, encephalopathy, tremor, and seizures, that are usually self-limiting, but rare lethal cases have been reported as well (117). Severe cases are often managed with corticosteroids, while tocilizumab is mostly ineffective in the treatment of ICANS, contrary to its effectiveness in CRS (116). Cross-talk between the engineered T cells and other immune cells, especially macrophages, can lead to the induction of systemic inflammation in the form of CRS which might cause leakiness of the blood brain barrier and symptoms of ICANS (116). Therefore, ICANS is often associated and correlates with the severity of CRS in patients, but it has also been reported in some cases in the absence of CRS (117).

On-target off-tumor toxicity is another challenge for adoptive T-cell therapies. In contrast to CRS and ICANS, this adverse event is not caused indirectly by associated endogenous immune cells but directly by the engineered T cells recognizing their cognate antigen or a cross-reactive antigen on healthy tissues. In particular, TCR-T cells warrant careful screening to avoid severe toxicity caused by autoreactivity due to lethal cases of tissue damage in the heart and brain in two early clinical trials targeting MAGE-A3 positive cancers (118, 119). As these TCRs were affinity enhanced, on-target off-tumor toxicity was likely increased. However, recent studies with affinity enhanced TCRs targeting MAGE-A4 showed clinical efficacy in the absence of severe TCR-T cell-mediated toxicity (120–122), indicating the necessity to investigate each single modified TCR for such risks although also non-modified TCRs have the potential for cross-reactivity. These cases highlight the difficulties of screening methods for TCR candidates to exclude potential common and individualized severe autoreactivity, which is highly difficult to be tested sufficiently at the preclinical level, but also indicate that TCRs have the potential to represent safe therapies with promising efficacy.

For the use of allogeneic CAR-T and TCR-T cells, a potential additional adverse event risk is allo-reactivity of allogeneic engineered T cells against foreign MHC molecules or minor histocompatibility antigens on host cells that can lead to GvHD. As described above, early clinical trials suggest that gene editing for the removal of the endogenous TCR is a viable strategy to avoid GvHD. Moreover, the use of allogeneic engineered T cells post-HCT, HLA-matching of the donor cells, and the use of non-αβ T cells are potential strategies (123). However, it remains to be seen how durable the persistence of allogeneic engineered T cells with these strategies is compared to their autologous counterparts, since elimination of the allogeneic T cells by the host immune system is a serious concern here.

Moreover, accompanying therapies, such as lymphodepletion prior to T-cell therapy, can add to hematological toxicities like cytopenia, which is a common adverse event with often unknown origin (26). Overall, this highlights the complex clinical landscape in regard to adverse events of CAR-T and TCR-T cell therapies and the need for better preclinical models to predict them early on (113).

Persistence of CD19 CAR-T cells for up to 10 years has been reported in two patients with chronic lymphocytic leukemia after remission (124). While CD8⁺ CAR-T cells were abundantly found in the initial response, it was almost exclusively CD4⁺ CAR-T cells that were present during long-term remission. However, poor T cell persistence has been reported in many CAR-T (125) and TCR-T cell (97) clinical trials against a variety of tumor entities and is often likely the reason for limited clinical efficacy or relapses. Therefore, administration of lymphodepleting regimens is commonly performed prior to the engineered T-cell therapy to increase T cell persistence (26).

Preclinical studies observed less exhaustion and improved antitumor activity of CAR-T cells with PD-1 knock-out (126, 127). Therefore, PD-1 knock-out is currently explored for the treatment of tumors with CAR-T (108, 125) and TCR-T cells (70) as a strategy to protect the engineered T cells from exhaustion and to enhance their persistence. Stadtmauer et al. used CRISPR-Cas9 genome editing to remove the endogenous TCR and PD-1 and introduced an engineered TCR specific for the cancer-testis antigens (CTAs) NY-ESO-1 and LAGE-1 to treat two patients with refractory melanoma and one with sarcoma. Engineered T cells trafficked to the sites of the tumor and reduction of the target antigens, likely as a response to the immune pressure of the TCR-T cells, was observed for both melanoma patients. Interestingly, no toxicity was observed and persistence of the TCR-T cells was increased in all three patients with at least 9 months compared to previous trials with T cells that retained their endogenous TCR and PD-1 expression. However, the number of patients in this first-in-human phase I clinical trial is low (NCT03399448) and expansion of the study is necessary. It also remains to be elucidated if the prolonged persistence is based on the ablation of PD-1 or in part on the removal of the endogenous TCR. Moreover, PD-1 ablation was also reported to cause increased functional exhaustion and cell death along greater activation in a CD19 CAR-T cell therapy (128). Results from knock-out experiments of PD-1 in mice with a chronic lymphocytic choriomeningitis virus (LCMV) infection showed that CD8⁺ T cell exhaustion can not only occur in the absence of PD-1 but PD-1 even protected the cells from overstimulation and terminal differentiation to an exhausted effector phenotype at the site of infection (129). Suggesting that PD-1 could be relevant to fine-tune T cell responses in certain environments, such as high antigen load as is the case in viral infections. In that context, transient blockade of PD-1 with ICI could be superior over PD-1 ablation but our understanding how PD-1 signaling modulates gene expression during T cell responses remains enigmatic and needs to be further elucidated. A recent study showed that genes associated with survival and proliferation are resistant to PD-1-mediated inhibition while effector functions are regulated by it based on the TCR signal strength (130). Due to the context-dependent functions of PD-1 signaling, it remains to be seen if PD-1 ablation of genetically engineered T cells is an effective way of improving T cell persistence depending on the tumor entity and antigen load.

Moreover, a number of costimulatory switch receptors have been reported to prevent exhaustion of genetically engineered T cells and might increase their persistence (131–133). Switch receptors consist of the extracellular portion of an inhibitory receptor (e.g. PD-1, TIGIT, TIM-3) and the intracellular signaling domain of a costimulatory receptor (e.g. CD28, 4-1BB). Reports from the preclinical studies are encouraging with improved antitumor activity and persistence of genetically engineered T cells armored with these switch receptors, however, these signaling axes are complicated in nature. Tipping the scale from exhaustion to activation and not just balancing it as is the case with anti-PD-1/PD-L1 ICI might lead to overstimulation and a dysfunctional T cell phenotype.

Tonic endogenous TCR signaling was also associated with improved persistence of CAR-T cells in recent studies (125, 134). Removal of the endogenous TCR and PD-1 in mesothelin (MPTK)-specific CAR-T cells for the treatment of solid tumors resulted in poor persistence of the TCR-deficient CAR-T cells beyond 6 weeks in a phase I clinical trial with 15 patients (NCT03545815 (125)). Surprisingly, it was the TCR-positive CAR-T cells that became the main fraction after infusion in three patients despite their rare presence in the infused cell product. The authors replicated these findings in mice and hypothesized that tonic TCR signaling plays a beneficial role in CAR-T cell persistence. Off note, this was in the scenario of low-level engraftment and might be different when using lymphodepletion to increase engraftment. In line with these observations, another study also observed reduced persistence for TCR-deficient CD19 CAR-T cells in animal models (134). The role of tonic TCR signaling for the longevity of CAR-T cells (19) needs to be further addressed especially in the context of allogenic CAR-T cell therapies, where the removal of the endogenous TCR is already common practice to prevent GvHD (108, 109). TCR-T cells with endogenous TCR replacement are likely not affected due to tonic signaling of the introduced TCR as indicated by the results from Statdmauer et al.

A common form of tumor resistance to ACT is tumor antigen escape by loss or downregulation of surface antigens or peptide-HLA complexes (135). Targeting multiple antigens by engineering T cells with a dual CAR or a tandem CAR could reduce the risk of tumor antigen escape ( Figure 4A ). Preliminary results from a phase I study (NCT03233854) with a CD19/CD22 bispecific CAR showed clinical efficacy but antigen escape in relapses was mostly observed for CD19 and not CD22 antigen, suggesting that there was less immune pressure of the CAR-T cells on the CD22 target (136). This is supported by the observation that CD22 scFV ligation in the bispecific CAR showed less cytokine secretion than for CD19 scFV. Another bispecific CD19/CD22 CAR that uses a tandem approach showed good efficacy in 6 patients with r/r B-ALL and observed one relapse of blast cells with loss of CD19 antigen and diminished CD22 expression 5 months after treatment (NCT03185494 (137)). The single construct approach for multi-specific CAR-T cells might affect the antigen binding capabilities based on the design of the linkers and it could be favorable to use a bi- or tricistronic design that expresses individual CAR molecules on the same cell to avoid these problems. This has been done for tricistronic CD19/CD20/CD22 tri-specific CAR-T cells that were able to target B-lineage ALL independent of CD19 expression in vitro and in animal models (138). Results from a phase I study (NCT03289455) with bicistronic CD19/CD22 bispecific CAR-T cells for the treatment of 15 patients with r/r B-ALL showed 86% CR and one-year overall survival and event-free survival of 60% and 32%, respectively (139). Dual targeting of B7-H3 and CD70, which are overexpressed in a variety of solid tumors, with a tandem CAR elicited superior tumor control and overall survival in a lung cancer and melanoma xenograft model (140). Moreover, infusing patients with a mixture of mono-specific CAR-T cells or TCR-T cells with different specificity could also be a viable option and has recently been done for patients with solid tumors that received up to three neoTCR-T cells with different specificity (97).

Multi-targeting of different antigens seems to be viable strategy to overcome tumor antigen escape and preliminary results suggest that it might increase antitumor immunity against tumor cells that co-express multiple antigens. However, this approach is limited so far by the number of known promising tumor antigens and will benefit from the discovery of additional tumor-associated and tumor specific antigens in the future.

The treatment of solid tumors is one of the most difficult areas in the field. Limited T-cell infiltration into the tumor as well as T cell exhaustion due to the immunosuppressive TME poses a high risk for an insufficient response to the treatment. In addition, the heterogeneity of antigen expression and the lack of truly tumor-specific surface antigens in solid tumors (112) can cause severe off-tumor toxicity in healthy tissues. Due to the difficulty of identifying tumor-specific surface antigens on solid tumors as targets for CAR-T cell therapies, strategies with higher specificity and less off-tumor toxicity have been developed, such as the synthetic Notch (synNotch) receptor designs ( Figure 4B ). Recognition of a primary antigen by the synNotch receptor cleaves an orthogonal transcription factor from the cytoplasmatic tail and induces the expression of a CAR that can recognize a secondary antigen on the tumor cell (141). T cell activation and tumor cell killing only occurs if both antigens are expressed on the tumor cell. This concept has been applied in a number of preclinical studies for solid tumor models and demonstrated improved specificity for the treatment of solid tumors (142, 143). SynNotch circuits can also be used to improve the specificity of engineered TCRs for selective killing of tumor cells, which has been demonstrated for a SynNotch-TCR against melanoma cells in vitro (144).

TCR-T cells are inherently equipped for the recognition of intracellular antigens through peptide-HLA presentation which opens up a treasure trove of tumor-associated or tumor-specific antigens that could be exploited for TCR-T cell therapy of solid tumors ( Figure 4C ). As mentioned above, a preliminary study showed the feasibility and efficacy of TCR-T cells in two melanoma and one sarcoma patient (70) and clinical trials for the treatment of MAGE-A4 positive solid tumors showed clinical efficacy in subset of patients with an ORR of 24% (9/38) (121). Foy et al. demonstrated recently the feasibility of TCR-T cell therapy targeting personalized neoantigens (97). A combinatorial screening approach with whole exome sequencing (WES) and RNA sequencing of the patients’ tumors was used to predict potential neoantigens. In a next step, multimeric labeled peptide-HLA complexes were generated for reactivity assessment of the predicted neoantigens in peripheral blood of patients and neoTCRs were identified and manufactured for 16 patients. Despite this impressive demonstration of feasibility of such a complex workflow in a clinical setting, efficacy and T cell persistence was limited. However, there are over a hundred clinical trials with adoptive TCR-T cell transfer registered in clinicaltrials.gov with the majority of them for the treatment of solid tumors (145). Hurdles for TCR-T cell therapy include its dependency on a specific HLA genotype, restricting it to a specific patient population in most cases, and its susceptibility to HLA-downregulation. Moreover, intracellular antigens can also be targeted with TCR-like CARs that use scFV molecules that recognize specific peptide-MHC complexes. Recently, a phase I clinical trial with a TCR-like CAR T-cell therapy targeting MAGE-A4 peptide-HLA-A*02:01 complexes for the treatment of solid tumors has been initiated (146) and evaluation of the clinical efficacy of more TCR-like CAR formats in clinical trials will be of great interest for the use of this concept. Overall, combinatorial approaches with a variety of different interventions will likely be necessary for the treatment of solid tumors in the future in order to overcome their complex mechanisms of immune evasion.