Mammals developed about 200 million years ago while a majority of bird species developed only about 66 million years ago [7]. Bird viruses thus had a relatively shorter time to adapt to the immune system of their hosts than viruses of mammals. Multicellular organisms, like birds and mammals can respond to virus infection, in particular by a type I interferon response (see below). As an avian virus, NDV has evolved viral immune escape mechanisms in birds. These interfere with the type I interferon mediated host response. Importantly, this viral escape mechanism is species specific and does not apply to non-permissive hosts.

NDV is an avian paramyxovirus type 1 (APMV-1). Such viruses have a negative sense single-stranded RNA (−ssRNA) as genome. Some strains show in non-permissive hosts a natural oncotropism (i.e., tumor selective viral replication), oncolytic potential and immune stimulatory properties.

The phylogenetic classification system of NDV has recently been updated [8]. NDV strains are classified according to their pathotypes and virulence as either lentogenic (low), mesogenic (medium) or velogenic (high). Velogenic strains are highly infectious in birds and are distinguished as viscerotropic or neurotropic pathotypes.

Genome sequences for many strains of NDV are available on the web at www.ncbi.nlm.nih.gov. All genome sizes of NDV obey to the rule of six which is characteristic for APMV-1 [9]. The genomic RNA contains a 3′-extragenic region known as leader and a 5′-extragenic region known as trailer. These are regions for control of virus transcription and replication and also for encapsidation of newly synthesized RNAs into virus particles. Leader and trailer flank the six genes (3′-N-P/V-M-F-HN-L-5′) of the viral genome. The genes code for nucleoprotein (NP), phosphoprotein (P), matrix protein (M), fusion protein (F), hemagglutinin-neuraminidase protein (HN) and large protein (L). The genome has a large capacity (>5 kb) for the incorporation of transgenes [4].

Infection of cells by NDV involves as a first step binding of the virus to the host cell’s surface via the cell-binding domain of the HN molecule [10]. α2,6-linked sialic acid was demonstrated to serve as high-affinity receptor for HN and for binding of NDV to cells [11]. Cell surface binding of the virus is followed by activation of the fusion protein F. A concerted action of HN and F leads to fusion of the viral and the host cell membrane so that the viral genome can enter the cytoplasm of the host cell. NDV can also utilize a clathrin-mediated endocytosis route and macropinocytosis as alternative endocytic pathway to enter cells [12]. In the cytoplasm, the −ssRNA is transcribed into messenger RNA. The viral mRNAs remain polyadenylated and carry 5′-phosphate groups (5′-PPP) in contrast to mammalian mRNAs which are either capped or contain base modifications. The viral mRNAs are then translated into the corresponding viral proteins.

The second step relates to viral replication using a newly produced nucleocapsid as positive-strand “anti-genome” [13]. The viral nucleocapsid consists of a single species of viral RNA (15,186 to 15,198 nucleotides in length) and replicon complex proteins (NP, P, L). Once produced, the nucleocapsid as “antigenome” is used as template for viral replication. The viral proteins P and L associate to form the enzyme RNA-dependent RNA polymerase, which is essential for viral replication. Viral replication can produce 1000 to 10,000 copies per cell. The newly produced viral genomes become encapsulated at the plasma membrane. A membrane budding process then enables the release of new virus particles.

Infection of normal cells from non-permissive hosts (e.g., mouse or man) usually does not proceed to the second step [14]. In NDV-infected non-activated or activated T cells, intracellular viral RNA (− and + strands and double-stranded (ds) RNA) can be detected within 1 to 12 h [14]. The viral replication cycle is prevented by interferon-induced genes (ISGs) such as myxovirus (influenza) resistance A (MxA), 2′-5′-oligoadenylate synthetase/ribonuclease L (OAS) and dsRNA-activated protein kinase R (PKR) (13). Mammalian and avian cells use OAS/RNase L to degrade cellular and viral RNA and retinoic acid inducible protein I (RIG-I) to enhance interferon induction as the first line of defense against viral infection. Viruses have developed diverse strategies to escape antiviral effects of OAS. NS1 protein of Influenza virus A acts upstream of its pathway while other viral proteins such as Theiler’s virus L protein act downstream.

NDV was recently reported to be able to establish persistent infection in tumor cells in vitro. There was a contribution of the cleavage site in F and of a second sialic acid binding site of HN. Persistent infection avoids oncolysis and immunogenic cell death (ICD) (see below) and facilitates viral spread from cell to cell as a potential mechanism to escape host antiviral responses [15].

OVs such as naturally occurring and attenuated NDV can be used alone or in combination with cancer vaccines for human cancer therapy [16]. OVs can also be used as vectors for gene therapy of cancer. Basic research in molecular biology and virology allowed to elucidate paramyxovirus replication and pathogenesis. Such studies were primarily performed with the rodent paramyxovirus sendai virus. Reverse genetics is a method by which infectious negative strand RNA viruses can be generated entirely from cloned DNA (c-DNA) [3,17,18,19]. This molecular engineering technology has allowed to develop recombinant oncolytic paramyxovirus strains with additional transgenes.

Harnessing OV-mediated anti-tumor immunity by therapeutic transgenes, pharmacological agents or bispecific antibodies has been the topic of a special issue of Frontiers in Oncology [1].

Type I interferon (IFN-I) responses serve as a first line of host defense against microbe invasion. Type I interferons can induce tumor cell apoptosis and anti-angiogenesis via signaling through the common type I interferon alpha receptor (IFNAR). IFN-I can also exert direct effects on cells of the immune system [20].

It is of great interest to compare the interferon response to NDV in permissive cells from birds as natural hosts to the response in cells from non-permissive hosts including humans. Infection of chickens with virulent NDV is associated with severe pathology, morbidity and mortality. Recently, immune responses were described of mature chicken bone-marrow-derived dendritic cells (BM-DC) upon infection with NDV strains of differing pathogenicity. Gene expression profiling revealed increased expression of melanoma differentiation associated gene 5 (MDA-5), of the helicase LGP2, the Toll-like receptors (TLR) 3 and 7, of type I interferon (IFN-α, IFN-β), IFN-γ, of interleukins (Ils) IL-1β, IL-6, IL-10, IL 12, IL-18, of chemokine ligand CCL5, and of major histocompatibility molecules MHC-I and MHC-II. Velogenic NDV showed a stronger replication capacity in BM-DCs than lentogenic NDV [21].

Of importance for the pathogenicity of NDV in birds is the capacity of certain viral products to antagonize the interferon response. In avian cells, NDV has developed a frameshift variant of the viral phosphoprotein P to escape type I IFN mediated anti-viral responses [22]. The incorporation of two G nucleotides at the RNA editing site of the P gene results in the frameshift variant V protein. Recently it was demonstrated that the interferon antagonistic activities of the V proteins of NDV correlated with their virulence [22]. The V protein interacts specifically with bird proteins [22]. It thereby can inhibit IFN signaling by targeting signal transducers and activators of transcription 1 (STAT1) for degradation. It can also interact with MDA-5, leading to the inhibition of interferon regulatory factor (IRF)-3 activation and IFN-β induction [23].

Upon virus infection, MDA-5 (or RIG-I in humans) recognizes ds viral RNA with 5′-triphosphate as being foreign and distinct from self-RNA. This initiates a strong type I interferon response [24,25]. Mitochondrial antiviral-signaling protein (MAVS) is an essential adaptor protein in RIG-I mediated antiviral innate immunity. Recently, a MAVS gene from goose (goMAVS) was identified. This bird derived MAVS mediated the activation of the type I interferon pathway in a species-specific manner [26]. Cells from birds can activate [26] or inhibit [23,27] a type I interferon response quite similar to cells from mammals. However, decisive protein–protein interactions in the interferon signaling pathway are bird-specific.

In mammalian cells, NDV induces a strong interferon response which involves an early and a late phase. This leads to inhibition of virus replication. The early phase response is initiated by cytoplasmic RIG-I-like receptors (RLR) upon recognition of foreign viral RNA while the late phase is initiated by membrane-bound IFNAR upon recognition of their respective ligands [20,27]. RLR is a family of cytoplasmic RNA helicases that includes RIG-I and MDA-5. These RNA sensors signal through the mitochondrial adaptor MAVS, recruiting kinases to activate the nuclear transcription factors NF-κB and IRF-3, and induce the transcription of IFN-I and proinflammatory cytokines.

In RIG-I-deficient fibroblasts, cytokine production is abrogated in response to NDV, sendai virus, vesicular stomatitis virus, hepatitis C virus or influenza A and B virus, thus demonstrating the importance of this foreign RNA sensing cytoplasmic receptor [28].

Other pathogen recognition receptors (PRR) are TLRs. These consist of more than 10 members and are expressed on the cell surface membrane or on endosomes. TLRs appear to be required to induce an interferon response in plasmacytoid dendritic cells (pDC) [29], while RLRs are critical to sense NDV by conventional myeloid DCs (mDC), macrophages, and fibroblasts. Details about the interferon signaling cascades and their regulation have been described [30,31]. Cytosolic nonself DNA recognition via the cyclic GMP-AMP synthase (cGAS) pathway also leads to IFN-I induction [32]. Upon sensing DNA viruses, RNA polymerase III is reverse transcribed to produce short RNAs, which are recognized by RIG-I. Latest updates on RIG-I also reveal long noncoding RNAs (LncRNAs) as being involved in regulation of RLR pathways. Antiviral IFN-I signaling is also regulated by covalent protein modification via ligation of small ubiquitin-like modifier (sumoylation) [32]. Not only foreign nucleic acids but also “unmasked”, misprocessed, or mislocalized host-derived RNA or DNA molecules can be recognized by RLRs or cGAS thus leading to proinflammatory or autoimmune disease [32].

During the late phase (8–18 h) of the IFN response [25], the induced type I IFN molecules (IFN-α and -β) secreted during the early phase, interact with the cell surface expressed IFNAR. These cytokine receptors are expressed by cells of all lineages, but not on mature erythrocytes. Nearly 20 four-helix bundle cytokines exist in humans and mice to interact as ligands with IFNAR. IFNAR consist of two chains, R1 and R2. The cytoplasmic domains of these chains are physically associated with Janus-family tyrosine kinases (JAKs) and Tyrosine-kinase (Tyk)-2. Ligand–receptor interaction initiates an amplification loop of the IFN response, which involves signaling via STAT proteins and IRF-7 [33].

Human mDCs were infected by NDV to study an uninhibited cellular response to virus infection [34]. The new approach integrated genome-wide expression kinetics and time-dependent promoter analysis. Interestingly, the anti-viral cell-state transition during the first 18 h post-infection could be explained by a single convergent regulatory network. A network of 24 transcription factors was predicted to regulate 779 of the 1351 up-regulated genes. It was concluded that the proposed network is effective in changing the cells underlying biology. The timing of this step-wise transcriptional signal propagation appeared as highly conserved. The effect of NDV on human DCs was also analyzed by the release of cytokines important for Th1 or Th2 polarization. Human monocyte-derived DCs were found to become polarized towards DC1 [35]. Signaling through RIG-I and IFNAR was found in further studies to be of importance for immune activation of DCs and other immune cells by NDV in mouse or human [25].

In contrast, signaling through RIG-I and IFNAR is successfully antagonized by respective inhibitory proteins from ebola virus (EBOV). EBOV, belonging to the family of filoviridae, was first discovered in 1976. It infects cells from primates (gorilla, shimpanzee, human). Primary target cells are macrophages and DCs. Zoonotic transmission of EBOV to humans causes severe and often lethal hemorrhagic fever. Disease characteristics are a systemic inflammatory response syndrome (SIRS), disseminated intra-vascular coagulation, systemic hemorrhage and multiple organ failure. Two of the 8 viral proteins of EBOV are involved in immunosuppression. In multiple ways they prevent type I interferon signaling. VP35 antagonizes the early phase of the interferon response. VP24 is an antagonist of the late phase. The molecular details of this virus-inhibited cellular response to infection have been described elsewhere [36]. Recently, a novel EBOV glycoprotein (GP) modified recombinant NDV (rNDV) was constructed as potential vaccine. The EBOV GP was expressed at a high level. Upon immunization, guinea pigs developed high levels of neutralizing GP-specific IgG and IgA antibodies [37].

3. Over 50 Years of Clinical NDV Application

3.1. 1960s to 1970s: Post-Operative Treatment with Oncolysate Vaccines

As early as 1965, Cassel and Garret from Atlanta (GA, USA) reported on oncolytic NDV (strain 73 T) as an antineoplastic agent [6]. Thereafter they observed the development of post-oncolytic immunity [132]. These authors were pioneers in developing NDV-based viral oncolysate vaccines for post-operative active-specific immunization of stage II malignant melanoma patients (two Phase II clinical studies involving 32 and 51 patients). A ten-year follow up of these 83 treated patients revealed that over 60% were alive and free of recurrent disease [133].

Later, a similar Phase II study with autologous NDV-modified tumor lysate vaccines has been conducted in Germany involving 208 patients with locally advanced renal cell carcinoma. Kirchner et al. concluded that the results demonstrated improved disease-free survival (DSF) in comparison with survival data published for similar patients who were treated by surgery alone [134].

3.2. 1990s to 2000s: Post-Operative Treatment with Autologous Live NDV-Modified Tumor Cell Vaccine (ATV-NDV)

To increase the immunogenicity of virus-modified tumor vaccines, a new concept was developed at the German Cancer Research Center (DKFZ) in Heidelberg, Germany. The idea was to develop an irradiated virus-infected autologous live cell tumor vaccine (ATV-NDV). The concept was tested with success in various metastatic animal tumor model systems [135,136]. After a series of tests with human tumor cell lines or with tumor cells from fresh operation specimens, human tumor cell modification by virus infection (NDV lentogenic strain Ulster) was found to be an efficient and safe way to produce a homologous human ATV-NDV cancer vaccine. This had pleiotropic immune stimulatory properties [137].

One clinical study evaluated the effect of vaccine quality parameters (i.e., cell number and cell viability) on the survival of primary operated breast cancer patients treated post-operatively with ATV-NDV [138]. Overall survival (OS), four years after surgery, was 96% for patients who received a high-quality vaccine (n = 32) compared with an OS of 68% for those who had received a low-quality vaccine (n = 31), a difference that was statistically significant [139].

A Phase II trial involved patients with locally advanced colorectal carcinoma (n = 57). Two years after surgery, OS for patients treated post-operatively with ATV-NDV was 98% compared to 74% for historical control subjects [138]. Using ATV-NDV vaccine obtained from cell culture, a further Phase II study was performed with patients suffering from glioblastoma multiforme. The median OS of vaccinated patients (n = 23) was 100 weeks versus 49 weeks in 87 non-vaccinated control subjects from the same clinic and the same time period (p < 0.001) [140].

A similar Phase II trial involved 20 patients with stage III and IV head and neck squamous cell carcinomas (HNSCC). The study demonstrated feasibility and safety of the ATV-NDV vaccine regimen and suggested prolongation of five-year survival [141].

Finally, a prospectively randomized Phase II/III trial (n = 50) was performed in stage IV colorectal carcinoma patients after resection for hepatic metastases. The objective was to investigate the efficiency of ATV-NDV as a tertiary prevention method. Evaluation was performed after an exceptionally long follow-up period of 9–10 years. While there was no significant difference between control and vaccinated rectal cancer, a significant benefit was seen in the colon cancer subgroup with regard to metastasis-free survival and overall survival: In the vaccinated arm, only 30.8% had died, while in the control arm 78.6% had died [142]. The trial provides evidence for the clinical value of the vaccine ATV-NDV. Its mechanism of function has been reviewed [143].

In 2003, a Phase III trial from China reported results from 310 patients suffering from Gastrointestinal Carcinoma (stage I-IV) which were treated with autologous tumor vaccine and NDV vaccine (strain La Sota). In comparison to 257 patients in which resection was performed without further vaccination, a significantly improved mean and median survival was observed for the vaccination group (7 years versus 4.46 years) [144].

3.3. 2000s: Systemic Application Studies with Oncolytic NDV

A selected case series study of four high grade GBM was performed in Hungary by Csatary and colleagues by i.v. application of an attenuated oncolytic veterinary vaccine strain, which was termed MTH-68/H. It was reported in 2004 on radiographically-documented responses and on long survival with improved symptomatology [145]. About ten years earlier, the same group had reported on a placebo-controlled Phase II clinical trial in which the virus was applied to patients (n = 26) with various advanced chemorefractory cancers. Interestingly, the virus had been applied at high doses via inhalation as a means to target lung metastases [146]. The high virus doses had been well tolerated. The 2-year survival rates had been 21% in the virus treatment group versus 0% in the placebo group. Although the study had not been randomized, the authors suggested a decrease in cancer-related symptoms and better survival [146].

Another case-series study in GBM was performed in Israel by Freeman, Zakay-Rones and colleagues with the lentogenic strain NDV-HUJ. The virus was administered i.v. using intra-patient dose-escalation followed by three cycles of 55 billion infectious units. Toxicity was minimal and a maximal tolerated dose was not reached. One patient experienced a complete response, although of only transient duration (about three months), while the others developed progressive disease [147].

The oncolytic NDV strain PV701 was applied in the United States by Wellstat Biologics in patients with advanced cancers that were unresponsive to standard therapy. This strain had been extensively tested in vitro [39] and in human tumor xenotransplanted mice [40]. Seventy-nine late-stage cancer patients were given escalating doses of virus i.v. Doses of 12 × 10⁹ to 12 × 10¹⁰ infectious particles (plaque forming units/m²) were well tolerated [148]. 3 Phase I trials were undertaken, involving 113 patients, to evaluate effects, such as virus dose, schedule and i.v. infusion rate. Adverse events were flu-like, tumor-site-specific or those occurring during infusion. When patients were desensitized with a lower initial dose, the maximal tolerated dose (MTD) could be increased 10-fold. In 95 evaluable patients, there were 10 responses (six major and four minor) with progression-free survival ranging from 4 to 31 months [149].

Another interesting activity of NDV relates to the clinically important process of liver fibrosis. Activated hepatic stellate cells (HSCs) represent a crucial factor in the development of liver fibrosis and are involved in the development of hepatocellular carcinoma (HCC). NDV was reported to be able to repress the activation of human HSCs and to be capable, upon systemic application, to revert the development of hepatic fibrosis in mice [150].

3.4. 2010s: Combining Oncolytic Virus Modified Vaccines (ATV-NDV) with Costimulatory Bispecific Antibodies

Most of the clinical trials performed with the vaccine ATV-NDV showed promising results with improvements in OS by about 30% of the treated patients. The remaining about 70% have to be considered as immunological non-responders. Since T cell anergy (non-responsiveness of TA-specific T cells) is a major problem in cancer patients and since this is often due to insufficient costimulation, a strategy was developed, as early as 1999, to augment T cell costimulatory signals in the vaccine ATV-NDV [151].

The strategy consisted of adding NDV-specific single chain antibodies with dual specificity (bispecific scFv antibodies, bsAb) to the vaccine ATV-NDV. In comparison to other bispecific antibodies, such as bispecific T cell engagers (BITES), which target TAs, the above bsAbs target viral antigens (VAs). They can thus be applied to any type of tumor cell modified by NDV. ATV-NDV tumor vaccine with attached anti-HN-anti-CD3 (αHN-αCD3) and anti-HN-anti-CD28 (αHN-αCD28) bsAb exerted in vitro, upon stimulation of allogeneic human peripheral blood mononuclear cells (PBMC), strong and durable antitumor effects against human tumor cell monolayers [152]. This anti-tumor activity was independent from recognition of TAs.

The activity of the bsAbs could be further augmented by introducing the cytokine IL-2 which binds to its high affinity receptor CD25 on T cells. For this purpose, a trispecific immunocytokine reagent (αHN-IL-2-αCD28) was constructed, produced and tested in vitro as above [153]. A transcriptome analysis and cytokine profiling of naive T cells stimulated by ATV-NDV tumor vaccine via attached αHN-αCD3 (for signal 1) and αHN-IL2 (for signal 2) revealed unsuspected costimulatory activity of the cytokine IL-2 [154]. The above trispecific immunocytokine, attached to ATV-NDV, provided the strongest T cell costimulatory activity in combination with a suboptimal amount of attached αHN-αCD3. This effect was most likely due to the concomitant transmission of signal 1 via CD3 and costimulatory signals 2a and 2b via the two T cell co-receptors CD28 and CD25 [152].

A Phase I clinical study evaluated the recombinant bispecific protein αHN-αCD28. In this autologous situation, the ATV-NDV vaccine provided TAs for signal 1 and αHN-αCD28 for signal 2. This dose-escalation study, in which the vaccine was modified by increasing amounts of the bsAb, involved 14 colorectal carcinoma patients with late-stage disease (stage IV with liver metastases). There were no severe adverse events. Before the vaccination, none of the patients had detectable levels of cancer-reactive blood circulatory T cells (ELISPOT test). Interestingly, after vaccination, all patients became positive in the ELISPOT assay, at least once during the course of 5 vaccinations. Furthermore, there was a dose-response relationship with the bsAb. A partial response of metastases was documented in 4 patients. The study suggests that the bsAb modified vaccine ATV-NDV-bsαHN-αCD28 is safe and can re-activate possibly anergic T cells from advanced-stage cancer [155]. A further potentiation of the costimulatory effect can be expected from a vaccine of the type ATV-NDV-αHN-IL-2-αCD28.

Bispecific antibodies and also trispecific immunocytokines have been proposed to have great potential in future for targeting the immune system against cancer [156]. While NDV infection of tumor cells could break T cell tolerance in vitro with human cells [89], it was not sufficient to have a similar effect in vivo in late-stage cancer patients. To overcome anergic T cells in this situation and to reactivate them required stronger costimulation as exemplified with the bispecific anti-CD28 antibody attached to ATV-NDV [155]. Not only the costimulatory signal can be augmented by bsAb but also signal 1. If the TA mediated signal 1 is very weak, it could be intensified by a suboptimal amount of bsαHN-αCD3. The ATV-NDV-bsαHN-αCD28 vaccine would be further modified by attachment of bsαHN-αCD3.

The concept of increasing T cell costimulatory signals is complementary to the concept of decreasing T cell inhibitory signals via checkpoint inhibitory antibodies. It is likely that the ratio of positive to negative signals will be decisive for the T cell response.

Table 3 contains a summary of the mentioned studies of clinical application of NDV.

Table 3. Clinical application of NDV (Part I).

1. Post-operative application of NDV oncolysate vaccines in Stage II melanoma (n = 83) by WA Cassel and DR Murray (1977) (oncolytic strain 73 T)

2. Systemic treatment of advanced chemorefractory cancers; Phase II placebo-controlled trial (n = 33 versus 26 controls) by LK Csatary and S Eckhardt (1993) (oncolytic strain MTH/68)

3. Systemic treatment of glioblastoma multiforme (GBM) (n = 14); Intra-patient dose escalation followed by 3 cycles of 55 billion infectious units by AI Freeman and Z Zakay-Rones (2006) (lentogenic strain HUJ)

4. Phase I dose-escalation trials of intravenous virus administration in patients with advanced solid cancers resistant to standard therapy (n = 113) by AL Pecora and RM Lorence (2002, 2003) (oncolytic strain PV71)

5. Post-operative treatment with the irradiated live tumor cell vaccine Autologous Live NDV-Modified Tumor Cell Vaccine (ATV-NDV): (i) Early breast cancer (n = 63), metastatic breast cancer (n = 27) and metastatic ovarian cancer (n = 31) by T Ahlert and V Schirrmacher (1997) (lentogenic strain Ulster) (ii) Phase II trial in patients with locally advanced colorectal carcinoma (CRC) (n = 57) by D Ockert and V Schirrmacher (1996) (lentogenic strain Ulster) (iii) Phase II trial in patients with GBM (n = 23 versus 87 non-vaccinated patients from the same clinic in the same time interval) by HH Steiner and C Herold-Mende (2004) (lentogenic strain Ulster) (iv) Phase I/II trial patients with Stage III and IV head and neck squamous cell carcinomas (HNSCC) (n = 20) by J Karcher and G Dyckhoff (2004) (lentogenic strain Ulster)

6. Prospectively randomized Phase II/III trial to investigate the efficiency of ATV-NDV vaccination after liver resection for hepatic metastases of CRC as a tertiary prevention method (n = 51) by T Schulze and PM Schlag (2009) (lentogenic strain Ulster)