Viral Replicon Systems and Their Biosafety Aspects

Abstract

Introduction:

Viral RNA replicons are self-amplifying RNA molecules generated by deleting genetic information of one or multiple structural proteins of wild-type viruses. Remaining viral RNA is used as such (naked replicon) or packaged into a viral replicon particle (VRP), whereby missing genes or proteins are supplied via production cells. Since replicons mostly originate from pathogenic wild-type viruses, careful risk consideration is crucial.

Methods:

A literature review was performed compiling information on potential biosafety risks of replicons originating from positive- and negative-sense single-stranded RNA viruses (except retroviruses).

Results:

For naked replicons, risk considerations included genome integration, persistence in host cells, generation of virus-like vesicles, and off-target effects. For VRP, the main risk consideration was formation of primary replication competent virus (RCV) as a result of recombination or complementation. To limit the risks, mostly measures aiming at reducing the likelihood of RCV formation have been described. Also, modifying viral proteins in such a way that they do not exhibit hazardous characteristics in the unlikely event of RCV formation has been reported.

Discussion and Conclusion:

Despite multiple approaches developed to reduce the likelihood of RCV formation, scientific uncertainty remains on the actual contribution of the measures and on limitations to test their effectiveness. In contrast, even though effectiveness of each individual measure is unclear, using multiple measures on different aspects of the system may create a solid barrier. Risk considerations identified in the current study can also be used to support risk group assignment of replicon constructs based on a purely synthetic design.

Introduction

Viral RNA replicons are defined as self-amplifying RNA molecules of viral origin and originate from single-stranded positive- or negative-sense RNA viruses. They are a form of viral vectors and are generated by deleting the genetic information of one or multiple structural proteins of a wild-type virus. The remaining viral RNA can be used as such and brought into the cells as a naked replicon by transfection or electroporation, or via synthetic nanoparticles. A schematic representation of a wild-type infection by an enveloped positive-sense RNA virus compared with the transfection of cells with a naked replicon is shown in Figure 1A and B, respectively.

Figure 1.

Schematic representation of a wild-type infection with an enveloped positive-sense RNA virus (A), compared with the transfection of cells with a viral replicon based on a positive-sense RNA virus (B). The figure does not aim at a complete representation of the gene expression strategies of all positive-sense RNA viruses and, for reader clarity, only shows a representative example. (A) Infection of a cell with a wild-type virus: (1) Attachment, entry, and uncoating of the virus. (2) Translation of vRNA (∼mRNA) into nonstructural proteins. (3) Genome replication and synthesis of additional viral RNAs coding for structural and nonstructural proteins. (4) Translation of RNA into structural proteins. (5) Assembly of viral proteins and RNA into a new viral particle. (6) Egress of the infectious virus particle. (B) Transfection of a cell with a naked replicon: (1) Transfection of RNA. (2) Translation of vRNA (∼mRNA) into nonstructural proteins. (3) Genome replication and synthesis of additional viral RNAs coding for nonstructural proteins. mRNA, messenger RNA; NS, nonstructural; S, structural; vRNA, viral RNA.

Alternatively, a viral replicon particle (VRP) can be generated by simultaneously providing (RNA coding for) the missing protein components to the cell, after which particles are formed which are able to infect a cell but are unable to subsequently form infectious virus particles due to the absence of the coding sequences for (part of) the structural proteins. Representative examples of the production of VRPs based on an enveloped positive-sense RNA virus using cells (stably) expressing structural proteins or using nonreplicating helper RNA coding for structural proteins are presented in Figure 2A and B, respectively. Finally, copy or complementary DNA (cDNA) can be delivered to the cells instead of viral RNA wherein it is converted into viral RNA. The latter approach is referred to as DNA-launched replicons (DREPs).

Figure 2.

Schematic example of the production of VRPs using cells (stably) expressing structural proteins (A), compared with the production of VRPs by means of transfecting cells with nonreplicating helper RNA coding for structural proteins (B). The figure does not aim at a complete representation of all VRP production systems and, for reader clarity, only shows two representative examples. VRP, viral replicon particle.

Due to their self-amplifying nature, RNA replicons have gained much attention in research and development as well as clinical applications. By removing genes coding for structural proteins and replacing them with transgenes of interest, they form ideal vectors to deliver genes to cells of interest or to help present specific proteins to the immune system. Since genome replication and transcription will occur in the replicon-containing cell, higher expression levels of the transgenes can be obtained in comparison with replication-defective vectors.

To work safely with viral RNA replicons, specific safety measures may be needed depending on local legislation, such as the legislation related to genetically modified organisms. To determine suitable safety measures, a thorough risk assessment is required. So far, limited information is available on the factors to consider for such risk assessment. While, on the one hand, the risk group of the parental virus can be considered, multiple genetic modifications, on the other hand, may warrant downscaling of the risk. The aims of the current review article were to provide an overview of replicon systems that are currently available or in development, to summarize their biosafety characteristics, and to recommend on an approach for future assessments and potential downscaling.

This review focuses on the application of, as well as the biosafety considerations for replicons originating from positive- and negative-sense single-stranded RNA viruses. Application of double-stranded RNA viruses, which so far has only been explored sporadically, as well as of single-stranded RNA viruses that use reverse transcriptase (i.e., retroviruses) remains outside the scope.

Overview of Single-Stranded RNA Viruses as Backbone for RNA Replicons

Single-stranded positive- and negative-sense RNA viruses have been associated with multiple diseases in vertebrates^* and are classified in various risk groups. Table 1 provides a more detailed overview of the different viral families and species from which replicons have been generated. Risk classification mentioned in the table is based on the assessments of multiple international databases including the Netherlands Commission on Genetic Modification (COGEM), the Belgian Service of Biosafety and Biotechnology, the Central Committee on Biological Safety, Germany, and the Federal Office for the Environment, Switzerland (FOEN).

Table 1.

Overview of host range, disease, and risk group classification of viruses (wild type and, if relevant, vaccine strain) from which replicons have been derived (nonlimitative list)

Classification	Family	Virus	Host range	Disease		Risk group
				Human	Animal
Positive-sense nonsegmented	Astroviridae	Human astrovirus	Humans		—	2a
	Caliciviridae	Norovirus (Norwalk virus)	Humans; rodents; felines; canines; sea lions; pigs; sheep; cattle; bats		—	2
	Coronaviridae	Middle East respiratory syndrome coronavirus	Humans; dromedary; camel; bats			3
		Kunjin virus	Humans; bats; civets			3
		Severe acute respiratory syndrome coronavirus 2	Humans; bats; minks; cats			3
	Flaviviridae	Bovine viral diarrhea virus	Even-toed ungulates	—		2
		Classical swine fever	Pigs	—		4
		Dengue virus	Humans; mosquitoes		—	3
		Hepatitis C virus	Humans; nonhuman primates; mosquitoes		—	2
		Japanese encephalitis virus	Humans; mosquitoes and vertebrate hosts, primarily pigs and wading birds			3
		Kunjin virus	Humans; mosquitoes; birds; horses			2b
		Niénokoué virus	Mosquitoes	—	—	1c
		Tick-borne encephalitis virus	Humans; ruminants; birds; rodents; horses; carnivores; ticks		—	3
		WNV	Humans; nonhuman primates; equines; dogs; cats; sheep; llamas; alpacas; alligators; birds; squirrel; chipmunk; rabbits			3
		Yellow fever virus	Humans			3
		Yellow fever virus—vaccine strain 17D				2
		Zika virus	Humans; monkeys; sheep; goats; horses; cows; ducks; rodents; bats; orangutans; carabaos; mosquitoes		—	3
	Hepeviridae	Hepatitis E virus	Humans; fish; amphibians; moose; kestrels		—	2
	Picornaviridae	Mengovirus	Humans; rodents; pigs; monkeys		—	2a
		Poliovirus	Humans		—	2
		Foot-and-mouth disease virus	Cloven-hoofed animals (including bovids)	—		4
		Enterovirus 71	Humans		—	2
		Coxsackie virus	Humans		—	2
		Hepatitis A virus	Humans; nonhuman primates		—	2
		Human rhinovirus	Humans; primates		—	2
	Togaviridae	Salmon pancreas disease virus	Atlantic salmon (farmed salmonid fish)	-		2
		Chikungunya virus	Humans; nonhuman primates; rodents; birds; mosquitoes			3
		SFV	Humans; mosquitoes; wild birds; rodents; domestic animals and nonhuman primates			2
		SINV	Humans; mammals; birds; insects; amphibians			2
		VEEV	Humans; horses; burros; bats; and terrestrial mammals			3
		VEEV—vaccine strain TC83				2
		Western equine encephalitis virus	Humans; reptiles; bats; pheasants; wild birds; mosquitoes; horses; dogs; rodents			3
		Eastern equine encephalitis virus	Humans; reptiles; bats; pheasants; wild birds; mosquitoes; horses; dogs; rodents			3
Negative-sense nonsegmented	Bornaviridae	Borna disease virus	Horses; ruminants; dogs; squirrels; foxes			2 H; 3 Aa
	Filoviridae	Ebola virus	Humans; chimpanzees; gorillas, baboons; duikers; rodents; shrew; bats			4
		Marburg virus	Humans; nonhuman primates			4
	Paramyxoviridae	Measles virus	Humans		—	2
		Parainfluenza virus	Humans; birds			2
		Nipah virus	Humans; pigs; bats			4
		Human metapneumovirus	Humans		—	2
		Respiratory syncytial virus	Humans; cattle	—		2
		Newcastle disease virus	Birds	—		3
		Newcastle disease virus—vaccine strain LaSota				2
		Sendai virus	Rodents (mice)	—		2
	Rhabdoviridae	VSV	Humans; horses; cattle; pigs; mules; sand flies; grasshoppers; rodents			3
Negative-sense segmented	Arenaviridae	Lassa virus	Humans; Mastomys natalensis (multimammate rat)			4
		Junin virus	Humans; rodents			4
		Tacaribe virus	Humans; bats	a		2
		Machupo virus	Rodents; humans; ticks; mosquitoes			4
	Hantaviridae	Hantaan virus	Humans; rodents		—	3a
	Nairoviridae	Crimean-Congo hemorrhagic fever virus	Humans; cattle; sheep; goats; birds; ticks			4
	Phenuiviridae	Huaiyangshan banyangvirus (also known as “severe fever with thrombocytopenia virus”)	Humans; goats; sheep; cattle; dogs; chickens, rodents, pigs; ticks			4
		Rift Valley fever virus	Humans; ruminants; camelids; mosquitoes			3
		Uukuniemi uukuvirus	Humans; ticks; blackbirds; mosquitoes	a		2

The table only refers to vaccine strains for which a risk group classification has been published by international databases.

^a Based on classification published by the Belgian Service of Biosafety and Biotechnology (SBB).¹³⁰

^b Based on classification published by the Federal Office for the Environment, Switzerland (FOEN).¹³¹

^c Based on classification published by the Central Committee on Biological Safety, Germany (ZKBS).¹³²

SFV, Semliki Forest virus; SINV, Sindbis virus; VEEV, Venezuelan equine encephalitis virus; VSV, vesicular stomatitis virus; WNV, West Nile virus.

Literature Survey to Identify Applications and Risk Considerations of Viral RNA Replicons

A literature review was undertaken in view of a project issued by the COGEM. The aim of the study was to compile scientific information related to the developments and use of replicons, as well as their potential risks. To formulate search strings, the following keywords were selected: Replicon/Vector, Virus/Replication competent/RCV, Biosafety/Risk/Adverse effect. After selecting keywords, a typical search string was composed using Boolean operators: (VIR* OR “REPLICATI* COMPET*”) AND (REPLICON OR VECTOR) AND (BIOSAFE* OR RISK OR “ADVERSE EFFECT” OR SAFE* OR ENVIRONMENT*) AND (RNA).

Relevant scientific publications from peer reviewed journals were retrieved using the databases Scopus and Web of Science. No language or publication restrictions were applied, and studies were not selected based on quality. The literature search was concluded on January 25, 2021, and resulted in ∼1400 publications. Additional publications were retrieved until April 30, 2021, and were based on reference lists of publications identified in the primary literature study, on reference lists in guidance documents and advices, and on internet searches using terms relevant for the current study.^†

Most publications presented applications of a previously generated construct, referring to a common original article in which that construct was initially described. While some described features that are claimed to be beneficial for biosafety, none had the intention to study biosafety.

Further insight into risk assessments was obtained through an analysis of guidance documents, advices, and recommendations issued by (inter)national organizations and institutions related to (bio)safety. These included but were not limited to the COGEM, the European Biosafety Association, the American Biosafety Association, the National Institutes of Health, the World Health Organization, the Canadian Centre for Veterinary Biologics, and the Central Committee on Biological Safety (Zentrale Kommission für die Biologische Sicherheit, ZKBS). A search was performed using the term “replicon” on the websites of the respective organizations and institutions.

Application of Viral RNA Replicons

In total, the application of 54 replicon systems based on a variety of single-stranded RNA viruses was identified, including 3 chimeric approaches, that is, replicons based on 2 viral species. A summary of applications of the different replicon systems is provided in Table 2. A more detailed description of replicon application is provided below.

Table 2.

Application of viral RNA replicons (nonlimitative overview)

Classification	Family	Virus (wild type or vaccine strain)	Study viral (genome) replication	Study of highly infectious viruses	Antiviral drug discovery	Cell modification in vitro	Vaccination against infectious disease	Gene therapy/anticancer therapy	References (nonlimitative list of [review] articles)
							(Pre-) clinical	(Pre-) clinical
Positive-sense nonsegmented	Astroviridae	Human astrovirus							92
	Caliciviridae	Norovirus (Norwalk virus)							7,12,93
	Coronaviridae	Middle East respiratory syndrome coronavirus							94
		Severe acute respiratory syndrome coronavirus							12,35,95
		Severe acute respiratory syndrome coronavirus 2							11,96
	Flaviviridae	Bovine viral diarrhea virus							17,97
		Classical swine fever							16–19
		Dengue virus							9,12,93
		Hepatitis C virus							12,98
		Japanese encephalitis virus							9,99,100
		Kunjin virus							3,18,93
		Niénokoué virus							9
		Tick-borne encephalitis virus							16
		WNV							12,17,101
		Yellow fever virus							9,12,18
		Zika virus							9,12
	Hepeviridae	Hepatitis E virus							4,12,102
	Picornaviridae	Mengovirus							24
		Poliovirus							1,12,27
		Foot-and-mouth disease virus							103
		Enterovirus 71							12,104
		Coxsackie virus							12
		Hepatitis A virus							12
		Human rhinovirus							12
	Togaviridae	Salmon pancreas disease virus							12,17
		Chikungunya virus							18,32
		SFV							16–18,31,32,105
		SINV							17,18,106
		VEEV							16–19,32
		Western equine encephalitis virus							18,93
		Eastern equine encephalitis virus							18
	Chimeric	SINV/VEEV							16,19
		SINV/Western equine encephalitis virus (Eastern equine encephalitis)							107
		Vesicular stomatitis virus/chikungunya virus							18
Negative-sense nonsegmented	Bornaviridae	Borna disease virus							108,109
	Filoviridae	Ebola virus							8,110
		Marburg virus							8,111
	Paramyxoviridae	Measles virus							13,18,32
		Parainfluenza virus							112
		Nipah virus							113
		Human metapneumovirus							114
		Respiratory syncytial virus							93,115
		Newcastle disease virus							116
		Sendai virus							23,36,117,118
	Rhabdoviridae	VSV							17,18,32,119
Negative-sense segmented	Arenaviridae	Lassa virus							8,18,120
		Junin virus							8
		Tacaribe virus							8
		Machupo virus							8
	Hantaviridae	Hantaan virus							121
	Nairoviridae	Crimean-Congo hemorrhagic fever virus							8
	Phenuiviridae	Huaiyangshan banyangvirus (also known as “severe fever with thrombocytopenia virus”)							8,122,123
		Rift Valley fever virus							17,124
		Uukuniemi uukuvirus							8

, Preclinical trials only; , clinical trials.

Replicons in Research

Study viral (genome) replication

Several viruses do not amplify efficiently in cell culture. To overcome the need to use wild-type virus particles to study genome replication, a self-replicating naked RNA replicon of the virus can be created and transfected in a suitable cell line. These systems have been developed for the following: poliovirus,¹ Semliki Forest virus (SFV),² Kunjin virus,³ hepatitis C virus,⁴ hepatitis E virus,^5,6 and norovirus.⁷ The resulting cell line, stably expressing the replicon RNA, can subsequently be used to study the effect of viral gene expression on intracellular events, as well as to study antiviral drugs, in particular substances that affect viral genome replication. In these cellular systems, no structural proteins are provided in trans, as they do not aim to generate VRPs.

Study of highly infectious viruses at lower containment level

Due to their high pathogenicity and the limitations in the prevention or treatment of infection, viruses of risk group 4 (RG4) can only be handled in BSL-4 containment conditions. To allow research on these highly pathogenic agents at a lower containment level, various replicon systems have been generated (reviewed by Hoenen et al.⁸). This is the case for multiple human RG4 viruses belonging to, for example, Arenaviridae, Phenuiviridae, Hantaviridae, Paramyxoviridae, and Filoviridae. For RG4 viruses of veterinary importance, RNA replicons (sometimes also referred to as minigenomes) have been designed for foot and mouth disease virus and classical swine fever virus.

Also for risk group 3 agents, such as for Flaviviridae (reviewed by Kummerer⁹) and, more recently for the Middle East severe respiratory syndrome coronavirus¹⁰ and the acute respiratory syndrome coronavirus 2,¹¹ RNA replicon systems have been developed to allow research at a lower containment level.

Antiviral drug discovery

A replicon-based assay allows for a high-throughput approach to test large libraries of compounds. In general, structural genes are replaced by reporter genes within the replicon RNA, after which cells are transfected with naked replicons or infected with a VRP to allow genome replication. In case structural proteins are provided in trans, also viral egress and a subsequent single cycle of infection will occur. A potential inhibitory effect of drug substances can be assessed by means of reduced expression of the reporter protein.

It remains outside the scope of this publication to describe each system in detail. For that, the reader is referred to scientific reviews, such as, but not limited to, Hoenen et al.⁸ and Fernandes et al.¹²

Cell modification in vitro

Viral RNA replicons have also been used for cell modification and differentiation in vitro. Replicons originating from Paramyxoviridae have been used to make induced pluripotent stem cells (iPSCs). In brief, a measles virus-based replicon has been generated to introduce reprogramming factors, such as OCT4, SOX2, KLF4, or cMYC in cells. Replicon genomes are subsequently eliminated from the derived iPSCs, taking their transient nature into account, leaving a virus-free cell system for research.¹³

Also, Sendai virus vectors have been shown to efficiently generate human iPSCs from human fibroblasts¹⁴ and human blood cells.¹⁵ Here, the issue was raised of the sustained cytoplasmic replication of the Sendai viral vectors after the iPSCs have been established. Several approaches have been explored to overcome this sustained replication, as discussed further.

Since replicons can modify cells but do not bear the risk of insertional mutagenesis (see the Genome Integration section), they are promising tools in the generation of stem and progenitor cells for future treatment of genetic and degenerative disorders.

Human and veterinary medicinal applications of replicons

Vaccination against infectious disease

Replicons can be exploited in several ways to prime the immune system. Their main advantage over conventional (i.e., nonreplicating) messenger RNA (mRNA) is that the antigen being expressed is proportional to the number of RNA transcripts, whereby conventional mRNA may require large doses or repeated administrations.

Administration of replicon RNA can be achieved as VRPs, liposome-encapsulated particles (e.g., liposomes, dendrimers), naked RNA replicons, or synthetic cDNA. Such vaccination has resulted in transient expression of the gene of interest, strong immune responses, and protection against viral challenge.

A detailed overview of all replicon systems used for vaccine development remains outside the scope of this report. For that, the reader is referred to scientific reviews, such as, but not limited to, Bloom et al.,¹⁶ Hikke and Pijlman,¹⁷ and Lundstrom.¹⁸ Several replicon systems intended to be used as vaccines against infectious disease are currently being explored in clinical trials (reviewed by Blakney et al.¹⁹).

Gene therapy and anticancer treatment

Replicons can also be used to deliver genes of interest in the context of gene therapy and anticancer treatment. Due to their cytoplasmic localization and self-replicating nature, significantly elevated and prolonged levels of transgene expression can be achieved without the risk of genome integration.

Most current applications are related to cancer treatment. For example, viral replicons can be used to deliver transgenes that, in turn, can affect tumor growth and/or attract cytolytic effector cells. This has been explored for an SFV-based DREP encoding human papillomavirus early proteins E6 and E7.²⁰ Using intradermal delivery followed by electroporation, it was demonstrated that the replicon induced effective, therapeutic antitumor immunity.²⁰ Also, injection of SFV-based replicons coding for anti-programmed death ligand 1 monoclonal antibodies²¹ or a Sendai or Kunjin virus-based replicon encoding mouse granulocyte–macrophage colony-stimulating factor^22,23 resulted in a potent antitumor response in in vivo models.

Finally, a mengovirus-based replicon has been used to modify tumor environment in the liver by inducing local recruitment of innate immunity effectors to the tumor, which complemented the effect of peptide-based vaccines in tumor regression.²⁴

In addition, replicons proved useful as delivery vehicles for small RNAs (sRNAs) that target specific promotor regions. They can induce long-term transcriptional gene regulation mechanisms, including transcriptional gene silencing or transcriptional gene activation, with subsequent long-term therapeutic effects for various diseases. Their nonintegrating nature and potential for inducing short-term expression is also of interest. Furthermore, genome editing could be achieved using RNA replicons with clustered regularly interspaced short palindromic repeat/CRISPR-associated protein 9 (CRISPR/Cas9). The application of replicons as delivery vehicle for sRNAs and CRISPR/Cas has been reviewed by Baltusnikas et al.^25,26

In view of cancer treatment, some replicon systems have intrinsic oncolytic properties. For example, RNA replicons derived from poliovirus showed a direct oncolytic effect on tumor cells of various origin in vitro based on their cytopathic effects.²⁷ The oncolytic effect was confirmed in vivo where it was additionally shown that the replicons not only exerted an effect at the site of inoculation but were also able to reach tumor cells that had metastasized from the initial site of implantation. Also, for vesicular stomatitis virus (VSV) replicons, an intrinsic oncolytic effect was demonstrated without vector-associated toxicities. The preferred genome replication in tumor cells, due to significantly attenuated antiviral responses in these cells, further supports the safety of the VSV replicon application.^28,29 Also, for recombinant SFV particles and naked viral RNA replicons, promising results have been obtained in view of cancer therapy because they ensure a high level of transgene expression, a rapid and strong cytopathic effect,³⁰ and an upregulation of interferon-stimulated genes.²¹

Several replicon systems intended to be used as anticancer therapy are currently being explored in clinical trials³¹ (reviewed by Blakney et al.¹⁹ and Lundstrom.³²).

Risk Considerations for Replicon Systems

This section focuses on specific considerations that may affect the risk of replicon systems for the user and/or the environment. While reviewing the different considerations, it must be emphasized that for each consideration, the relevance for the risk assessment will depend on the type of replicon system and its application. Additionally, it is noted that this review focuses on the hazards of the replicon system itself. When transgenes are inserted in the replicon RNA, these transgenes and their expression products may additionally affect the hazards of the system (e.g., oncogenes, toxins). The assessment of the transgene, however, is not part of the current review.

Naked Replicons

Naked replicons comprise the genetic information for nonstructural proteins and, if relevant, for some of the structural proteins in case they contain important elements necessary for efficient RNA replication and/or other sequences required for nonstructural protein or transgene expression. Consequently, transfection of cells with naked RNA allows for genome replication but not for generation of replicon particles.

Genome integration

Upon cell infection, some viruses integrate their genome into the host chromosome, either as part of their life cycle (e.g., retroviruses) or accidentally. Genome integration may not only potentially promote long-term persistence of a virus in the cell, but it could also lead to drastic consequences for the host cell, such as gene disruption, insertional mutagenesis, and/or cell death. For DNA viruses, the genome forms a potential substrate for host genome integration, without the need for prior processing. However, most RNA viruses are not able to deliberately integrate their genome into the host chromosome, as their genetic information resides in RNA molecules and not in DNA.

The only exception is retroviruses because they possess the ability to transcribe their viral RNA genome into DNA, a mandatory step for productive retroviral infection. For all other RNA viruses, genome integration has not been reported or has been reported only as an incidental event.³³ No data were found on the actual consequences of such rare integration events. Furthermore, no data were found on the likelihood for and the potential consequences of integration of replicons based on cDNA (DREP). Whether or not the reverse transcriptase encoded by the second open reading frame (ORF) of the human Long Interspersed Nuclear Element 1 in eukaryotic cells could allow for genome integration of replicons remains to be studied.³⁴

Persistence in the host cell

Several RNA viruses have developed specific mechanisms, for example, at the level of viral transcription and replication, that could lead to the establishment of persistent infection. To follow-up viral genome replication and to screen drug compounds in vitro, this feature has been exploited in RNA replicon systems based on, among others, hepatitis C virus,⁴ severe acute respiratory syndrome coronavirus,³⁵ norovirus,⁷ and hepatitis E virus.⁵ In brief, cells are stably transfected with (cDNA for) replicon RNA carrying a drug resistance gene, after which RNA genome replication is maintained in the cell.

For other applications, for example, when cells will be used for additional applications such as stem cell therapy, or when subsequent activities must be performed with less containment measures, prolonged presence of the RNA replicon is undesired. In these cases, conditional replication of the RNA replicon provides a means to control replicon expression. Ban et al.³⁶ described modified Sendai virus replicon systems in which various temperature-sensitive mutations were introduced. When transduced cells were maintained at permissive temperature (maximum 37°C), the temperature-sensitive Sendai virus replicon was detected on 3 dpi as based on a green fluorescent protein (GFP) signal. However, switching to nonpermissive temperatures (38°C or 39°C) resulted in a disappearance of the GFP signal. Moreover, when placing the cells again at 37°C, GFP was not detected in the cells by which the authors concluded that transgene-free human iPSCs were obtained.

Also, the use of micro (mi)RNA or small interfering (si)RNA provides an interesting means to express conditionally replicon RNA. For example, a Sendai virus-based replicon was extended with a sequence encoding a target sequence for the so-called miRNA 302. This miRNA is specifically expressed in pluripotent stem cells and targets the RNA of the polymerase L gene. As a consequence, the replicon system is expected to disappear from the reprogrammed cells. However, to conclude on the effectiveness of the approach, it is advised to check the cells for the absence of syncytia by microscopy, the absence of Sendai virus nucleoprotein (NP) expression by immunofluorescence, and the absence of Sendai virus NP mRNA by quantitative polymerase chain reaction (PCR).³⁷

To control replicon replication, also the use of riboswitches has been studied. In brief, a riboswitch is a regulatory segment of an mRNA molecule that binds a specific small molecule such as a therapeutic compound, resulting in a change in production of the proteins encoded by the RNA. For example, for alphaviruses, Bell et al.³⁸ incorporated a riboswitch into the replicon 3′ untranslated region (UTR). By doing so, they could successfully regulate the expression of DNA-launched and VRP-packaged replicons. When integrating riboswitches into the 3′ and 5′ UTR of the subgenomic RNA region of the TC-83 strain of Venezuelan equine encephalitis virus (VEEV), replication could be regulated at an even greater fold.

The results clearly indicate that riboswitches hold promise for the development of novel vaccination strategies, whereby gene expression can be controlled in such a way that only one injection will allow for priming and boosting of the immune response. Also, for gene therapy or cancer treatment, regulated expression of replicons will have clear benefits.³⁹ Further study will, however, be needed on the safety limits of each of the approaches, as mi- or siRNAs may not completely prevent the expression of all genes they act on.

Rather than including an element for conditional replication in the replicon system itself, external control factors can also be used. For example, cells can be treated with specific small molecules as was shown for cells transduced with a Borna disease virus-based replicon. Treatment with a small molecule, T-705, completely eliminated the replicon in persistently transduced cells.⁴⁰

Generation of virus-like vesicles

As mentioned above, the risk of the transgene remains outside the scope of this study. However, using viral envelope genes as the transgene in a naked replicon is relevant for this report since it may allow for the formation of the so-called virus-like vesicles (VLVs).^‡ In brief, upon their expression, the envelope proteins support the formation of vesicles in which the replicon RNA is packaged and spread from cell to cell. This phenomenon has been described when using a transgene coding for the G protein of VSV, a member of the Rhabdoviridae, in VEEV-based replicons or SFV-based replicons (Togaviridae).

VLVs with RNA replicon genomes were generated that could spread throughout a cell culture.^41–43 The majority of the infectious particles were smaller and less dense than either VEEV or SFV particles. Characterization by electron microscopy showed membrane-enveloped vesicles that contained the VSV-G protein. Infectious particles were apparently generated by budding of vesicles containing VSV-G protein and the RNA replicon.⁴¹ Similarly, VLVs were generated when an envelope protein of murine leukemia virus was expressed by an SFV replicon.⁴¹ From these studies, it is concluded that the formation of VLVs cannot be excluded when the coding sequence for (homologous or heterologous) envelope proteins is used as transgene.⁴⁴

Effects on nontargeted cells

Replicons could potentially exert negative effects on cells that are not targeted for the delivery of the gene of interest. To limit the replicon genome replication and transgene expression to targeted cells or tissues, different approaches have been described. One such approach employs the use of miRNAs, which are involved in silencing gene expression. The miRNAs bind to their cognate target RNA that possesses a miRNA-binding sequence generally referred to as the miRNA recognition element (MRE). Subsequently, the miRNA will silence the target RNA. Lee et al.⁴⁵ used this process to engineer a dengue virus-based replicon that could not replicate in the liver by inserting a liver-specific MRE into the viral mRNA. Binding of liver-specific miRNA with the MRE incorporated in the replicon RNA conferred an inhibitory effect on the replicon RNA.

Viral Replicon Particles

The risk considerations as mentioned for naked replicons are also relevant for VRP. The sections below elaborate on additional risk considerations that must be taken into account when producing and/or applying VRP.

Generation of replication competent viruses during production: “primary” replication competent virus

In the creation or production of VRP, there is a stage in which different viral genetic elements come together to assemble the infectious particle. In that stage, there is a potential risk for recombination due to homologous sequences, with the formation of “primary” replication competent virus (RCV). For nonsegmented single-stranded RNA viruses, recombination may theoretically occur between replicon RNA and co-replicating helper RNA (recombination is even less likely in case of nonreplicating helper RNA as it will only be translated but not transcribed or replicated).

For segmented, single-stranded negative-sense RNA viruses, recombination may occur between the transcription plasmids and the expression plasmids when homologous sequences are present, although the event appears to be rare (reviewed by Han and Worobey⁴⁶). Depending on the virus family, the likelihood for recombination may differ. Based on that, different strategies have been developed to reduce the likelihood of such recombination, as explained below. The value of these strategies has already been described for other RNA viral vector systems, the most comprehensive for lentiviral vectors (reviewed by Pauwels et al.⁴⁷).

Using bipartite helper RNA system

For those replicon systems where structural genes are provided via helper RNA, such as, but not limited to, those generated from Togaviridae, the likelihood of recombination can be reduced by dividing the genetic elements coding for the structural proteins over multiple helper RNAs. The rationale is that increasing the number of helper RNAs increases the number of recombination events that are required to generate a full functional viral genome. This is clearly illustrated in a study by Pushko et al.⁴⁸ where a monopartite and a bipartite helper RNA system were used for a VEEV replicon. In the monopartite system, where complementing genes were present on only one helper RNA, plaque formation indicative for recombination was found for different types of helper RNAs used, even when only very limited sequence homology was present with the replicon RNA. However, when using the bipartite system where the complementing genes were divided over two helper RNAs, no infectious virus was detected in the plaque assay.

Similar observations were made for Sindbis virus (SINV), where no recombinants were detected using one helper RNA coding for the capsid protein and a second helper RNA coding for viral membrane proteins.⁴⁹ These results were further confirmed in later studies by Hyvärinen et al.⁵⁰ It is highlighted that in a bipartite system, single recombination events may still occur, but these will not result in plaque-forming virus since only capsid or membrane proteins will be included.⁴⁸

Beissert et al.⁵¹ developed a novel bipartite vector system using trans-amplifying RNA. In brief, an alphavirus replicon encoding a particular transgene is used, but rather than including all nonstructural genes of the alphavirus, the genetic sequence encoding replicase is deleted from the replicon. Replicase activity is subsequently provided in trans by an optimized nonreplicating mRNA (nrRNA). In case the replicon could be complemented with genes encoding structural proteins, it is highly unlikely that additional recombination with the nrRNA would occur as recombination with nrRNA is highly infrequent.⁵² In consequence, a productive infection is equally unlikely to occur.

Blakney et al.⁵³ developed a system called splitzicon. Hereby, the genes encoding the nonstructural proteins and the gene of interest are located on separate RNA molecules but still exhibit the self-amplification properties of replicon RNA.

Limiting sequence homology between genetic elements

Whether or not recombination will occur, at least in part depends on the amount of sequence homology between the genetic elements. As a rule of thumb, the higher the homology between sequences, the higher the likelihood for recombination. Therefore, limiting sequence homology will reduce the likelihood of recombination. Depending on the replicon system, limitation of sequence homology can be achieved at different levels. Representative examples are described below.

For replicons based on VEEV, as an example of positive-sense single-stranded RNA viruses, it was shown that the rate of recombination in a monopartite helper RNA system was inversely related to the size of deletion in the nonstructural protein genes in the helper RNA.⁴⁸ However, even when the helper RNA contained <1 kb of nonstructural gene sequence, recombinants were detected, as evidenced by the generation of plaques in a plaque assay. The authors suggested that the likelihood for recombination was not only determined by the amount of sequence homology but also that specific sequences in the helper RNA may have affected the rate of nonhomologous recombination independent of deletion size, although the sequences of the recombinants were not determined.

Sequence homology can also be diminished by introducing mutations in coding sequences for the structural genes. In this perspective, Widman et al.⁵⁴ used a VEEV replicon to supplement the genetic information for the C protein in a West Nile virus (WNV) genome containing a 70-codon deletion in the C protein. With the aim of producing VRPs but avoiding recombination between the modified WNV genome and the VEEV replicon, 36 silent mutations were incorporated in the C gene, thereby generating a sequence that was no longer complementary to the WNV genome and thus preventing replication of a recombinant genome.

For negative-sense, single-stranded segmented RNA viruses such as the Phenuiviridae, recombination at the level of the UTR is a specific risk. Their UTRs direct replication and transcription of viral RNA are sufficient to allow encapsidation of viral RNA into ribonucleoprotein complexes.^55–57 Their terminal panhandle sequences are conserved between the different genome segments.⁵⁸ In the context of replicons, it is to be considered that, when segments previously and deliberately made devoid of their UTR recapture a homologous or heterologous UTR by recombination, an autonomously replicating virus particle may be the result.

For example, in a Rift Valley fever virus-based replicon system where an expression plasmid coding for the ORF of the M segment (no M-UTR) is co-transfected with transcription plasmids coding for the S or L segment, recombination may allow for the M ORF to obtain an UTR of the S or L segment and consequently allow replication and encapsidation of the M segment. The above-mentioned recombination events have never been described in nature, considering that the likelihood for a single crossover in two RNA segments is extremely low. Moreover, in the unlikely and mainly theoretical event that such recombination would occur, the resulting virus that harbors two genome segments with the same UTR is likely attenuated, as was demonstrated in experimental studies using reverse genetics for Bunyamwera virus.⁵⁸

Avoiding packaging of genetic elements coding for structural proteins in VRPs

Depending on the virus family, different elements are involved to ensure that all genetic elements are packaged into new virus particles, such as, but likely not limited to, specific packaging signals (reviewed by Mendes and Kuhn⁵⁹ and Twarock and Stockley⁶⁰), specific RNA–protein interactions (reviewed by Hornak et al.⁶¹), or UTRs.⁵⁷ During infection with a wild-type virus, these elements ensure that the various viral components are included in the virus particle, thereby releasing a fully infectious virus able to enter and replicate in new host cells. As a biological containment measure for replicons, it is reasoned that when the elements are removed or modified, the related RNAs will not be packaged into the VRP, and, consequently, the VRP itself will not harbor these RNAs and thus cannot express the encoded proteins. Depending on the replicon system, interference with the elements involved in packaging can be achieved at different levels. Representative examples are described below.

For those replicon systems where structural genes are provided via a co-replicating helper RNA, the likelihood of packaging of the helper RNA into the VRP can be reduced by not providing a packaging signal in the co-replicating helper RNA. That modifying packaging signals may result in hampered packaging has been demonstrated for RNA originating from SFV. Liljeström and Garoff² described defective helper RNA, derived from SFV RNA, that lacked a large portion of the nonstructural region, including the region corresponding to the packaging signal identified for SINV.

This defective helper RNA was able to package SFV replicons but was not packaged itself. Similar results were obtained by White et al.⁶² However, for replicons generated from SINV, it was shown that even a co-replicating helper RNA lacking a packaging sequence is still co-packaged at a rate of about 1/300 of that of single RNA containing a packaging sequence.^63,64 Of particular interest is the C protein. Packaging of the co-replicating helper RNA encoding the C protein without the packaging signal may, at least in part, be the result of the presence of other sequences in the C protein that support packaging.⁶³ Even when using a C protein from a different alphavirus, formation of particles containing the co-replicating helper RNA could not be prevented.⁶³

As explained above, for negative-sense, single-stranded segmented RNA viruses such as the Phenuiviridae, it is particularly relevant to remove the UTRs of the genome segments that are not to be packaged into the VRP, taking into account that the presence of the UTR is sufficient to allow encapsidation of that genome segment into ribonucleoprotein complexes.⁵⁷

Reducing the hazard of potentially formed RCV

While the previous approaches address the likelihood of recombination, safety of the replicon can be enhanced by modifying the nonstructural and/or structural proteins in such a way that in the unlikely event an RCV is formed, it does not exhibit hazardous characteristics.

For example, for SFV-based replicons, mutations were introduced that result in removal of protease activity of the capsid protein, so that in case of recombination of helper and replicon RNAs, the newly formed viral particle remains replication defective.⁶⁵ Also for SFV-based replicons, mutations were introduced in the cleavage site of the precursor protein p62.⁶⁶ The rationale was to construct a cleavage-deficient variant of the precursor protein that would not be cleaved into envelope proteins without external addition of chymotrypsin and thus would render the particles noninfectious unless chymotrypsin is externally added. The resulting RCV indeed turned out to be largely noninfectious, due to reduced binding and uptake into endosomes as well as the impaired ability to induce membrane fusion in the endosome. Based on these results, the authors concluded that a p62 cleavage-deficient helper function could be used to enhance the biosafety of the SFV replicon system.

Apart from introducing mutations in the structural proteins, an alternative approach is the use of structural proteins of a less pathogenic virus homologue. Provided that the less pathogenic phenotype is associated with attenuating mutations in genes encoding structural proteins, the VRPs generated are likely to display the biological properties of that virus, rather than that of the more pathogenic virus. For a VEEV-based replicon, Pushko et al.⁴⁸ described the use of helper glycoprotein genes derived from V3014, a recombinant VEE vaccine strain, which contain two strongly attenuating mutations. V3014 is avirulent upon subcutaneous inoculation of mice⁶⁷ and horses (Smith et al., unpublished observations), and therefore, any recombinant generated during VRP packaging is very likely to be at least as attenuated as V3014.

Similarly, structural genes of less pathogenic heterologous viruses can be used, thereby generating chimeric viruses. An example is a VEEV-based replicon system in which structural proteins of SINV were provided in trans.^68,69 In comparison to VEEV, SINV is less pathogenic to humans. Also, despite extensive use, no laboratory-acquired infections with SINV have been described that could be assigned to airborne transmission, whereas this has been described for VEEV.^70,71

Finally, also the genetic sequence of nonstructural genes can be modified to ensure that RCV, if formed, are not hazardous. Such was described for a Rift Valley fever virus system in which the coding sequence for the nonstructural protein in the S genome segment (i.e. NSs) was replaced by the enhanced GFP gene.⁷² Since NSs is involved in suppressing the host innate immune responses,^73–75 its absence could avoid those immunosuppressive effects.

Conditional generation of VRPs

As mentioned for the naked replicons, the production of VRPs can be made conditional, as illustrated below.

Harvey et al.⁷⁶ generated a stable baby hamster kidney packaging cell line, carrying a Kunjin virus structural gene cassette and a Tet-Off system. In the presence of tetracycline, no structural proteins were produced. Withdrawal of tetracycline from the medium resulted in the production of Kunjin virus structural proteins that were capable of packaging transfected and self-amplified Kunjin virus replicon RNA into secreted virus-like particles. The same cells were also capable of packaging replicon RNA from closely (WNV) and distantly (dengue virus 2) related flaviviruses.

For alphavirus replicons, it was shown that by incorporating miRNA-specific target sequences (tissue-specific MRE) into the helper RNAs, VRPs were found to be efficiently produced if miRNA-specific inhibitors were introduced into cells. However, in the absence of such inhibitors, cellular miRNAs were capable of downregulating helper RNA replication and consequently inhibited the VRP production.⁷⁷

Testing for the absence of “primary” RCV

Formation of RCV during production of VRP is a major risk consideration. Still, only limited data were identified on assays and/or limits of detection to provide evidence for successfully avoiding the presence of RCV (Table 3). In brief, data on RCV testing were identified for 5 of 34 identified replicon systems based on positive-sense single-stranded RNA viruses (including 1 chimeric system), 1 of 14 identified replicon systems based on nonsegmented, negative-sense single-stranded RNA viruses, and 1 of 10 identified replicon systems based on segmented, negative-sense single-stranded RNA viruses.

Table 3.

Examples of assays for the verification of replication competent virus presence

Classification	Family	Virus (wild type or vaccine strain)	Assay for RCV	Results	Detection limit	References
Positive-sense nonsegmented	Flaviviridae	Kunjin virus	Microscopy (detection of E protein by immunofluorescence during 2 passages in Vero cells)	No fluorescence detected	Not described	76
			Intracranial inoculation of suckling mice	No clinical signs upon inoculation with 4 × 106 IU	Not described
	Picornaviridae	Poliovirus	Intraspinal or intracranial inoculation of mice	No neurovirulence	Not described	125,126
	Togaviridae	SFV	Microscopy (infect BHK-21 cells and titrate supernatant in a PFU assay)	No PFU found in 4.6 × 109 particles from 13 independent SFV-lacZ recombinant stocks (frequency of recombination <4.6 × 109)	Not described	127
		VEEV	Microscopy (detection of syncytia upon 2 passages in Vero cells)	No RCV detected	1 RCV/1 × 1010 gm-replicon particles	128
			Microscopy (detection of syncytia upon 2 passages in Vero cells)	No RCV detected	<100 RCV/1 × 108 gm-replicon particles	48
	Chimeric	SINV/VEEV	Microscopy (detection of syncytia upon 5 passages in BHK cells)	No RCV detected	Not described	68
			Reporter assay (test supernatant of 5 passages in BHK cells for β-galactosidase expression)	No RCV detected in 40 batches	Not described
Negative-sense nonsegmented	Paramyxoviridae	Sendai virus	Microscopy (detection of syncytia)	Absence of syncytia	Not described	37
			qPCR (detection of NP mRNA)	Negative	2–3 Positive cells
			Fluorescence microscopy (detection of fluorescent-labeled antibodies to NP)	No fluorescence detected	Not described
Negative-sense segmented	Phenuiviridae	Rift Valley fever virus	Microscopy (detection of fluorescence during 2 passages in BHK cells)	No fluorescence detected upon inoculation of cells at an m.o.i. of 1	Not described	72
			Inoculation of suckling mice	No clinical signs upon inoculation with 1.0 × 104 TCID50	Not described	129

BHK, baby hamster kidney; E, envelop; gm, genetically modified; IU, infectious units; m.o.i., multiplicity of infection; mRNA, messenger RNA; NP, nucleoprotein; PFU, plaque-forming units; qPCR, quantitative polymerase chain reaction; RCV, replication competent virus; TCID, tissue culture infectious dose.

The amount of data is rather scarce compared with those for replication-deficient viruses, such as retro/lentiviruses and adeno-associated viruses, for which multiple guidelines have been developed. In brief, the FDA Guidance for Industry entitled “Testing of Retroviral Vector-Based Human Gene Therapy Products for Replication Competent Retrovirus During Product Manufacture and Patient Follow-up” provides recommendations regarding the testing for RCV during the manufacture of retroviral vector-based gene therapy products, and during follow-up monitoring of patients who have received retroviral vector-based gene therapy products. Recommendations include the identification and amount of material to be tested as well as general testing methods.

The requirement for testing adenovirus vectors for the presence of replication competent adenovirus (RCA) is described in Ph. Eur. 5.14 and the FDA guidance document entitled “Chemistry, Manufacturing, and Control (CMC) Information for Human Gene Therapy Investigational New Drug Applications (INDs)—Guidance for Industry” (2020). No acceptance criteria are specified in Ph. Eur. 5.14, but the FDA guidance document and the FDA BRMAC meeting number 30 (Adenovirus Titer Measurements and RCA levels, 2001) clearly state that the test is reported as “no RCA detected” if it complies with the acceptance criterion of <1 RCA/3 × 10¹⁰ virus particles.

Generation of VLVs

When using sequences of certain viral envelope proteins in helper RNA (or cDNA), generation of VLVs is a risk consideration during the production of VRPs, similar as described for the naked replicons (see section on Generation of virus-like vesicles).

Generation of RCVs During Application: “Secondary” RCV

Secondary events could lead to the generation of RCV, more specifically by recombination with or complementation by related viruses and/or unrelated viruses present in the host in vitro or in vivo during application of replicons.

In nature, recombination events do not appear to be a hallmark of negative-sense single-stranded RNA viruses.⁴⁶ While sporadic authentic examples indicate that homologous recombination does occur, recombination seems to be generally rare or even absent in most negative-sense RNA viruses, and most of the homologous recombination events reported in the literature were likely generated artificially. Natural recombination has, however, been described for positive-sense single-stranded RNA viruses, where it is an important mechanism for promoting genetic variation. For example, for flaviviruses, recombination has been described under natural circumstances for dengue virus.^78–80

Western equine encephalitis virus is believed to be a naturally occurring recombinant between SINV and Eastern equine encephalitis virus.⁸¹ Also, natural recombination among enteroviruses^82,83 and among coronaviruses⁸⁴ is frequently observed. Hereby, recombination of enteroviruses, in particular poliovirus, contributed to the emergence of pathogenic, circulating vaccine-derived polioviruses that have been complicating the World Health Organization's program for the global eradication of poliomyelitis,⁸³ whereas recombination of coronaviruses may have contributed to the emergence of the pandemic severe acute respiratory syndrome coronavirus 2.⁸⁵

Experimental conditions using cellular systems have confirmed and provided further insights in the recombination events. However, it is important to note that only one publication was identified in which recombination events were mimicked for viral RNA replicons.⁸⁶ The authors described a system based on self-replicating subgenomic RNAs (replicons) derived from tick-borne encephalitis virus, WNV, and Japanese encephalitis virus for investigating the ability of the flavivirus genomes to recombine. They showed that, even though the replicon pairs shared ∼600 nucleotides of identical sequence where a precise homologous crossover event would have yielded a wild-type genome, recombination was not observed.

Only two aberrant recombination events were detected, both of which yielded unnatural genomes containing duplications. Moreover, infectious clones of these genomes yielded viruses with impaired growth properties. For tick-borne encephalitis virus and WNV, their systems did not yield any viable recombinant genomes at all. These data indicate that even in the case of high sequence homology, recombination was a rare event and resulted in impaired viruses only. The impaired character of chimeric viruses was also demonstrated for Togaviridae.^87–89

Taking the above into account, recombination between replicon and (un)related viruses cannot be excluded. However, for recombination to happen, several conditions must be fulfilled. For example, replicon RNA and wild-type virus must be present in the same host cells. Therefore, if replicon and wild-type viruses enter different host cells, either because of differences in tropism or differences in route of application (e.g., injection in view of vaccination vs. natural infection), it is unlikely that recombination or complementation will occur. The likelihood of recombination or complementation is even further diminished if genome replication of the RNA replicon takes part in different cellular compartments compared with the wild-type virus replication.

Finally, even if replicon RNA and wild-type virus can infect the same host cell and replicate in the same cellular compartment, a phenomenon called superinfection resistance (also referred to as trans-inhibition or homologous interference) may prevent replication of the wild-type virus if the cell is already infected by the replicon. The phenomenon has been described for various single-stranded RNA viruses that are used for viral replicon design, including poliovirus,¹ alphaviruses,⁹⁰ measles virus, and hepatitis C virus. The mechanism of exclusion has been observed at various steps of the viral life cycle, including attachment, entry, viral genomic replication, transcription, and exocytosis.

Even though the data above indicate that the likelihood of recombination with or complementation by related and/or unrelated viruses is low, it cannot be excluded. Therefore, precautionary measures must be implemented mainly aiming at minimizing the risk of contact between replicon-containing materials or individuals on the one hand and the related and/or unrelated viruses on the other hand. In brief, measures can include, but are not necessarily limited to, selection of cells, experimental animals, and/or clinical trial subjects free of the related and/or unrelated viruses as tested by means of serological testing and/or PCR.

Conclusions

RNA replicons have gained much attention in research and development as well as clinical applications, such as gene therapy and vaccination. Despite these differences in origin and delivery mechanism, some common elements have been identified in the risk assessment for replicons as schematically represented in Figure 3.

Figure 3.

Summary of considerations justifying downscaling of the containment requirements of an activity with replicons. RCV, replication competent virus; RG, risk group.

In the case of VRP, the coding sequences for the structural proteins and those for the nonstructural proteins are separated from each other. The resulting VRP can enter the target cell as the wild-type virus would. However, while replication will occur, no new particles will be released from the cell. The risk classification of VRP can be downscaled substantially since they have turned into single round expression systems without pathogenic characteristics. Their classification is hardly related to that of the original virus. Nevertheless, uncertainty may still limit radical downscaling.

First, during the production of VRP when different viral genetic elements come together to assemble the infectious particle, there may be a theoretical or known possibility that recombination leads to the reconstitution of an RCV. Different approaches are discussed in this review for reducing the risk for primary RCV reconstitution. However, it should be noted that many considerations are of a theoretical nature. For example, different modifications in helper RNAs are expected to reduce the likelihood of recombination. Nonetheless, the exact impact of each modification (e.g., partial or complete deletion of structural genes, no sequence homology vs. limited sequence homology), as well as their (potential) cumulative nature (e.g., multiple plasmids and limited sequence homology) is not clear. Moreover, the (potential) impact is likely to differ between various virus families.

Second, although recombination leading to primary RCV is theoretically possible, there is little evidence that this recombination occurs at a significant frequency. Testing can provide further insights in the likelihood of recombination and the efficiency of the strategies for reducing the recombination potential. Still, each test method has its limitations. Robust data on which test to perform and what detection limits to consider are still restricted to only a few of the viral species for which replicons have been designed. One may question whether downscaling of the containment level for activities in which VRPs are produced can be allowed without confirmatory testing for the absence of RCV. Conversely, it can be questioned whether to perform repetitive testing once it has been established that RCV formation is unlikely.

Third, while replicons provide a safer approach for working with systems that are related to high-risk pathogens (in particular those belonging to RG4), the awareness of the high risk of the recipient may influence the willingness to downscale. Risk assessments are defined as an evaluation of likelihood and consequence. Even if the likelihood is low that recombination leads to a primary RCV, the possible consequences of handling a reconstituted RG4 virus at an inadequate containment level must be fully appreciated.

In the case of naked replicons, there is no phase during which the coding sequences for the structural genes are present. Primary recombination is therefore excluded. When solely using a coding sequence for nonstructural genes in the absence of any complementing sequences or sequences that allow for the formation of VLVs, the risk of the replicon seems negligible. Independent of the consideration of any inserted sequence, the risk group of the naked replicon can be downscaled as it lacks all pathogenic features. Although these elements provide a general argumentation, the specific nature of the replicon system (e.g., based on positive-sense vs. negative-sense RNA, use of co-replicating vs. nonreplicating helper RNA, segmented vs. nonsegmented RNA) may have to be used to refine the downscaling effort (e.g., different elements are involved in packaging for positive-sense compared with negative-sense RNA viruses, co-replicating helper RNAs are not used for negative-sense segmented RNA viruses).

While above we have highlighted the argumentation for downscaling the risk group of replicons, it must also be stressed that the situation differs depending on the type of activities performed with the replicon. For the worker handling the replicons, exposure is unintentional and unlikely even under low-level containment measures. In case of unintentional contact with the replicon, potential risks include persistence, VLV generation, effects of nontargeted cells, and for VRP production also secondary RCV formation. When replicons are intentionally administered in vivo (e.g., deliberate administration in view of vaccination of animals or delivery of genes for the treatment of human diseases), other risks may occur such as environmental risks.

For example, the participation of professional human research subjects in other studies besides that involving a replicon needs to be considered carefully, in particular when other studies involve vectors that may provide complementing sequences to the replicon. However, as replicons allow for a single infection only and do not have the capacity to cause a productive infection, the likelihood of recombination as a result of a consecutive study with a (potentially) complementing viral vector seems extremely low. Similarly, as the administration of the replicon will not result in viral replication and spread, shedding of replicons, if any, will be limited in time and space. Therefore, risk considerations are in a first instance relevant for and limited to the target cells. Cells or organisms treated with replicons can be handled in lower or no containment level once absence of replicon is demonstrated.

While secondary RCV formation based on recombination with a related or unrelated virus co-present in the host is a theoretical concern, its occurrence in reality remains unproven. The likelihood can be further reduced by avoiding simultaneous presence of the related or unrelated virus.

In conclusion, the most important concern is the primary production of RCV. In particular is the step where all genetic elements are present in the same cell. The recognition of this importance is illustrated by the fact that the majority of the publications address different strategies to reduce the risks for the formation of primary RCV. However, data on the actual need and/or efficiency of such measures remain scarcely documented. Even though it is not fully clear what the impact is of each of the measures, one might argue that the cumulative effect of different measures increases the safety profile.

In his “Swiss cheese model,” Reason⁹¹ visualized incidents as the result of the accumulation of multiple failures in defenses (represented as the holes in slices of cheese) that unfortunately align, creating an open “hazard trajectory” that results in harm. As long as enough layers of defense are in place, at least one with nonaligning holes, the probability of an accident can be minimized. For the risk measures of replicon systems, a similar reasoning could be proposed. Even though each individual measure is not yet proven leak tight, using multiple measures on different aspects of the system creates a solid barrier.

In the risk assessments summarized in this review, the starting point has been the recipient virus. The combination of safety measures allows a downscaling of the risk group as well as the containment measures. It is not clear if this approach, so embedded in the GMO legislation, is and will remain adequate. Design and application of replicons is a rapidly evolving field. Chimeric forms and even fully designed elements instead of those based on existing viruses are possible next steps. In such case, reference to a risk group determined for the recipient organism will not be possible, rather the risk classification will have to be developed de novo. The risk considerations identified in the current review can then be used to build up a “Swiss cheese model” that would support such risk group assignment, starting at the lowest level that would be applicable for self-amplifying RNA, that is, a naked replicon without a transgene.

Authors' Disclosure Statement

This article is based on research funded by the Netherlands Commission on Genetic Modification (COGEM).

Funding Information

No funding was received for this article.