## Introduction

A human analogue of insulin (Humulin) was the first biologic introduced into the market in 1983. Since then, biologics have been increasingly used as pharmaceutical agents, outpacing the market of small-molecule drugs1 with an estimated market size of around US$400 billion by 2025 (ref.2). Biologics are a class of medications that are derived from a biological source and typically fall into one of four major categories: monoclonal antibodies (mAbs), receptor modulators, enzyme modulators and vaccines3. Vaccines have been crucial in eliminating infectious diseases, such as smallpox4 and rinderpest in 2011 (ref.5), and in eradicating once-deadly diseases, such as diphtheria, measles, polio and rubella, by 99%, with many other diseases, such as mumps, pertussis and tetanus, nearing that stage. In the USA alone, it is estimated that in just one generation of children, 13 different vaccines have prevented up to 20 million diseases and 40,000 deaths, with economic savings of up to US$69 billion4. On a global scale, these numbers are much higher; one estimate shows that ten vaccines deployed in 94 low- and middle-income countries led to a US$586 billion reduction in illness-related costs, and a reduction of up to US$1.53 trillion when considering additional economic benefits6.

Vaccines are usually manufactured using cell-based expression systems, such as mammalian and insect cell lines, bacterial or yeast cultures. Plant molecular farming, that is, using plant cells or plants as expression platforms, is a rapidly emerging alternative, which was originally introduced in 1986 for the production of human growth hormone (HGH) in transgenic tobacco and sunflowers7. Plant molecular farming is currently used for the production of seasonal influenza vaccines8 and for Elelyso for Gaucher disease in the USA9. Although plant molecular farming may not replace industrial expression systems, it may have a unique role in the generation of vaccines in low-resource areas, for targeting niche and orphan vaccines, and for the production of virus nanoparticles (VNPs) for vaccine applications10. Furthermore, the rise of the coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has focused research attention on novel approaches to vaccines. Traditional vaccines are either live attenuated (that is, less virulent forms of the original pathogen), inactivated (that is, inactivated pathogen without disease-producing capacity), or recombinant proteins and viral vectors. In addition, the two vaccines that were first granted emergency authorization in the USA for COVID-19 are messenger RNA (mRNA) vaccines delivered in lipid nanoparticles11.

In this Review, we discuss challenges in current vaccine development and distribution, such as cold chain disruptions and lack of health-care professionals, which are particularly problematic in low-resource areas. Cold chain requirements pose a general logistical challenge for vaccines, as highlighted by the current rollout of the mRNA-based COVID-19 vaccines12, and traditional administration techniques, such as injections, require trained health-care professionals. Here, we investigate how molecular farming and nanotechnology-based strategies could provide an alternative strategy to traditional vaccines, enabling rapid development, effective deployment and safe administration of vaccines.

## Vaccine technology

### A brief history of vaccines

The history of modern vaccination began with Edward Jenner’s ‘live attenuated’ smallpox vaccine in the late eighteenth century13. Nearly a century later, in 1885, Louis Pasteur developed a rabies vaccine by drying the brains of infected rabbits; however, this vaccine was rather unpredictable and often caused serious side effects14. Shortly thereafter it was found that heat or chemical treatment could be used to inactivate bacteria, and vaccines for typhoid (Salmonella typhi), plague (Yersinia pestis) and cholera (Vibrio cholerae) were developed13. Vaccine production was greatly improved in the 1950s, when the development of cell culture techniques enabled the in vitro production of non-virulent viruses instead of requiring to be isolated from infected animals15. Bacterially expressed hepatitis B surface antigen then became the first vaccine produced by recombinant DNA technology16, which laid the foundation for the four major categories of modern vaccines: live attenuated, killed/inactivated whole organisms, subunit and toxoid vaccines (Table 1). Since then, new vaccine types have emerged, such as viral vectors and nucleic acid vaccines17. Today, 85 human vaccines and vaccine combinations are on the market.

### Challenges and innovations

Materials design can greatly improve vaccination approaches by enabling stabilization and controlled release, and/or by providing delivery devices for self-administration, which is particularly important for vaccines against viruses with a great degree of variability, such as the human immunodeficiency viruses (HIV-1 and HIV-2). HIV-1 shows worldwide variability as well as direct infection and destruction of immune cells. In addition, there is no validated animal model available for HIV-1 research, and the high rate of mutation allows the virus to escape antibody responses to targeted antigens18. High variability may be addressed by universal vaccines, which are effective against all virus subtypes and are impervious to future mutations by the pathogen. For example, the Gardasil-9 vaccine provides immunity against nine human papillomavirus (HPV) types, which should prevent >90% of HPV infections and subsequent cervical cancer19; however, expanding immunity protection to all known HPVs would enable HPV eradication. A universal vaccine against influenza would also greatly decrease infection rates, health-care costs and associated mortality, while improving patient compliance20. Nanotechnology-based approaches, such as HPV-encapsidating microneedles21 and slow-release implant technologies22, could further improve immunization against these difficult pathogens.

Vaccines are also being developed for non-infectious pathophysiological conditions, such as autoimmune diseases and cancers23. These vaccines are difficult to produce, because the molecular causes and antigens are often specific to the individual patient. Therefore, unlike for infectious disease vaccines, here it may be more advantageous to produce patient-specific vaccines — a considerable challenge for conventional production methods. Improvements in biotechnology methods, such as whole-exome or RNA sequencing24, may allow the isolation of patient-specific antigens for the creation of specialized vaccines.

Nanotechnology-based approaches have also played a key part in the development of COVID-19 mRNA vaccines, although it remains to be seen whether mRNA vaccines will provide long- or short-lasting immunity against SARS-CoV-2. Furthermore, the lack of thermostability of lipid nanoparticle formulations requiring storage and distribution in ultralow-temperature freezers presents a great challenge for vaccine distribution to rural and low-resource regions because these vaccines are temperature-stable only when frozen at –20 to –80 °C. Viral vector vaccines may be better suited for vaccination in remote areas, because they can be stored at higher temperatures of up to 4 °C. However, early reports suggest that viral vector vaccines do not reduce symptomatic infection as well as mRNA vaccines25, although the differing trial protocols and demographics make it difficult to compare the trials equally.

Humulin, the first biologic used in humans1, is expressed in Escherichia coli26. Advances in DNA technology have further enabled vaccine production in yeast, for example, for the production of subunit vaccines against the hepatitis B virus (HBV)27. In addition, mammalian cells, insect cells and chicken eggs can be applied as expression systems for vaccine production.

Bacteria are the most widely used expression system, benefiting from low cost, rapid growth and ease of use; in ideal conditions, E.coli can double its biomass within 20 minutes28. An estimated 30% of all biopharmaceuticals are currently engineered in bacterial systems, with E.coli being the most widely used strain29. Bacterial systems are also applied to generate DNA plasmids, which are subsequently used to produce the final biologic. The main disadvantage of bacterial systems is the lack of eukaryotic post-translational modifications. Without proper post-translational modification, purified proteins may behave differently in vivo26 compared with their non-recombinant counterparts, which can lead to diminished or total loss of activity, reduced half-life and decreased stability and/or immunogenicity30. Moreover, the presence of rare codons in eukaryotic genes can be problematic, because they can cause early termination during bacterial protein production, necessitating the redesign of genes31.

To overcome the problems associated with bacterial protein production, mammalian cell lines can be used. From 2016 to 2018, 84% of pharmaceutical proteins were made in mammalian cell lines, predominantly in Chinese hamster ovary (CHO) cells32. CHO cells achieve high protein yield (up to 10 g l–1 for some proteins)33, they can grow in suspension34 and they can withstand changes in external factors, such as temperature and pH35. However, although post-translational modifications in CHO cells more closely resemble those of human cells, they are not identical and can induce immune or other adverse reactions in patients36.

Alternatively, insect cells replicate faster than mammalian cells, enabling faster protein expression, but are more costly than bacteria owing to the requirement of specialized culture media28. A main disadvantage of insect cells is also the difference in post-translational modifications, as compared with human cells, which can make the biologics immunogenic37. Furthermore, some insects carry pathogenic viruses (for example, arboviruses) and, thus, insect cell lines must be closely examined before regulatory approval38. Furthermore, insect cells can produce proteases in response to viral transfection that can digest the protein of interest37.

Finally, chicken eggs are widely used for vaccine production, for example, for influenza vaccines39. Although well established, antigenic drift in chicken eggs can decrease vaccine efficacy, compared with vaccines produced in cell-based systems40. In addition, vaccine production in chicken eggs can take up to 6 months, whereas insect cell systems can produce vaccines in 6–8 weeks39. Possible contamination with human pathogens can also be a problem, which can be avoided by molecular farming41.

## Molecular farming

Since the production of HGH in transgenic tobacco7, many proof-of-concept and efficacy tests have been performed of plant-made therapeutics and vaccines for humans and animals42 (Table 2). A tobacco-produced mAb used in the production of a hepatitis B subunit vaccine43 and a Newcastle disease subunit vaccine for poultry made in cultured tobacco cells were the first plant-made recombinant proteins to receive regulatory approval in 2006 (ref.44) (Fig. 1). However, only one human therapeutic (Elelyso, a mitochondrial enzyme deficit therapy for Gaucher disease) produced by molecular farming has thus far been licensed by the US Food and Drug Administration (FDA)9. In addition, the first phase III human clinical trial for a plant-produced vaccine has just been successfully concluded. This virus like particle (VLP)-based quadrivalent seasonal influenza virus vaccine is currently undergoing final consideration for licensure in Canada8. The fact that only a few plant-produced therapeutic products have been clinically translated thus far may be more related to industrial and regulatory inertia than to product inadequacy, especially given the plentiful evidence of functional equivalency.

Transgenic plant and plant cell production of proteins laid the foundation of plant molecular farming; however, it was the advent of transient expression technologies that unleashed its true potential, including plant virus-derived vectors, somatic transformation of cells in normal plants (usually by infiltration of Nicotiana benthamiana with Agrobacterium suspensions), and hybrid methods involving delivery of replicating vectors by Agrobacterium42 (Box 1). These transient technologies enable protein expression within days of gene cloning, instead of relying on the generation of stable transgenic plants, which can take years. Moreover, yields are usually higher and more consistent, than transgenic plant systems, because these techniques do not depend on genome insertion sites and expression is not downregulated by methylation or other gene modifications, enabling much faster experimentation and optimization of expression. In addition, production scale-up is more efficient than transgenic plants and conventional cell-based technologies. Only a small volume of recombinant Agrobacterium suspension has to be grown for the biologic of interest, which can then be combined by agroinfiltration with as many plants — grown cheaply in nearby facilities — as required. It is striking that Medicago Inc. received funding from the US Defense Advanced Research Projects Agency (DARPA) Blue Angel initiative to produce ten million doses of current good manufacturing practice (cGMP)-level influenza vaccines 1 month after receiving the sequence of the virus45. Medicago Inc. has further announced preliminary success in investigating a VLP-based COVID-19 vaccine candidate, which was produced in just over 20 days after receiving the spike protein gene sequence46. Phase I clinical trial results showed ten times more production of neutralizing antibodies than in convalescent sera47. This vaccine candidate has recently entered phase II/III clinical trials and is administered along with GlaxoSmithKline’s AS03 adjuvant48.

### Algal cell expression

The microalga Chlamydomonas reinhardtii is the most commonly used algal system for molecular farming49 and has been applied to express mAbs (against glycoprotein D of the herpes simplex virus and human fibronectin type III), subunit vaccines (viral protein (VP) 1 of foot-and-mouth disease), allergens (peanut allergens), growth factors (vascular endothelial growth factor), and immunotoxins (chimeric antibody to CD22 and exotoxin A)50. Although the genomes of the mitochondria, nucleus and chloroplast of C.reinhardtii have been sequenced51, expression has been mainly applied in chloroplasts thus far, because the chloroplast genome is small and simple50, and biologics can be sequestered in the chloroplast without harming the host cells or being degraded51. However, similar to bacterial expression, post-translational modifications are lacking in this system. Alternatively, Lemna and moss systems have been explored for biologic expression52,53.

### Plants and plant cells

Using plants for vaccine production has the advantage that plants are cheaper than mammalian systems in terms of biomass production, and they have similar protein folding, assembly and glycosylation54. Therefore, plants can be applied to efficiently produce a range of pharmaceuticals at low cost. It is estimated that the final cost of producing biologics in plants is 68% of that in conventional production systems (site and batch costs are lower in plants, whereas downstream processing costs are identical)55. Furthermore, post-translational modifications, such as glycosylation, can be manipulated in plants. Thus, plant-produced products have appropriate glycan structures, which can improve half-lives of products and ease downstream processing56. For example, Elelyso is considered a ‘biobetter’ (biologics that improve upon existing biologics), compared with its CHO-produced counterpart, because the end product contains terminal mannose residues that aid in macrophage receptor binding57. In CHO cell production, mannose residues are attached in vitro, adding an extra step and increasing the costs58. Influenza virus haemagglutinin proteins made in plants were shown not to induce problematic adverse effects in clinical trials despite non-mammalian glycosylation59. In proteins that require human glycosylation, such as the HIV-1 Env gp140 protein, the yield can be substantially increased and the glycosylation pattern of HIV-1 Env gp140 can be altered through coexpression of human chaperonins in N.benthamiana plants, engineered to not produce xylose or fucose transferases60.

Plants and plant cells are also inherently safer than bacterial and mammalian systems owing to the low possibility of harmful contamination. In contrast to these systems, plants do not produce endotoxins and cannot be infected by pathogens harmful to humans61. Therefore, plants could potentially be used in edible vaccines62, although concerns of cross-contamination and accidental vaccination of wildlife have stalled development in this field.

Various vaccine candidates have already been produced by molecular farming, including HIV-1 Env60, ten different HPV L1 proteins63, West Nile virus envelope protein64 and dengue virus VLPs65. Notably, this list includes three different envelope glycoproteins or their derivatives. In addition, a recombinant double-stranded DNA (dsDNA) molecule has been encapsidated in HPV-16 pseudovirions, which were made in plants66, proving that DNA vaccines and their delivery systems can be succesfully produced in plants. Similarly, mRNA vaccines can be produced in plants; for example, recombinant mRNA molecules can be expressed in plants from DNA vectors and specifically encapsidated by adding an assembly signal in tobacco mosaic virus (TMV) coat proteins. Synthetic mRNAs made in vitro can also be encapsidated with purified TMV coat proteins, which was first demonstrated in 1956 with TMV RNA10. Encapsulation of target and self-replicating mRNA can be achieved in plants through co-expression of the desired mRNA and the TMV coat proteins67, making this an attractive and scalable approach towards plant-produced mRNA vaccines.

The limited commercialization of plant cell culture-based expression systems may be related to the lower protein yield (usually 0.01–10 mg l–1, although up to 247 mg l1 have been reported)68, compared with bacterial and mammalian cell culture, which can achieve up to 10 g l–1 (ref.33). Additionally, controlling the expression levels of certain proteins remains difficult, especially in transgenic plants. Expression can vary between different generations of plants, is highly dependent on the plant type, and can even fluctuate within the same plant’s tissues and organs42. Plants may also suffer from variability of environmental conditions, such as droughts and extreme heat — although this is less problematic in growth houses.

The cookie technique, that is, the transient transformation of plant cells to produce high-value proteins, may revolutionize the small-scale production of recombinant proteins and other biologics in cultured plant cells69. Here, cell packs or ‘cookies’ are made on porous supports by filtration of suspended cells, followed by incubation with Agrobacterium suspensions without resuspension. The bacteria are then washed out with growth medium, leaving behind a mass with many air spaces. The treatment results in transformation rates of up to 100%, compared with around 1% from co-incubation of cells with Agrobacterium in suspension70. Cell packs can be incubated for days in situ and can produce up to 47 mg kg–1 of recombinant protein. The technique is scalable from microlitre and millilitre volumes up to large-volume preparative columns used for filtration or chromatography, and thus can be used for high-throughput and low-volume screening and for subsequent production of trial batches of proteins.

### Plant molecular farming to overcome vaccination challenges

Plants can be used to produce vaccines at small scale and low cost in regions where vaccines are most needed, particularly in areas where it may be difficult to set up cGMP-level bacterial and mammalian cell culture systems. In addition, contaminants from plant molecular farming are not as harmful as endotoxins produced in bacterial expression61. Moving vaccine manufacturing on-site would also reduce cold chain requirements, because long-distance shipping would not be needed; this has been especially highlighted during the SARS-CoV-2 vaccine rollout. In 67 low-income countries, an estimated 90% of the population may not gain access to any SARS-CoV-2 vaccine in 2021 (ref.107). Therefore, plant molecular farming could reduce vaccine inequity and immunization costs108. Furthermore, production of vaccines by plant molecular farming does not require extensive space; for example, it was calculated that Italy could reach a 60% vaccination rate with 12,500 square metres of greenhouses45. Moreover, molecular farming is easily scalable, compared with fermentation systems, which must be adapted and optimized for scale-up, and which require costly bioreactors. By contrast, each plant is a bioreactor, and scale-up is simply achieved by growing more plants.

Plant molecular farming is particularly useful for producing plant viruses as nanocarriers for vaccines. Plant VNPs are very stable and can withstand temperatures outside cold chain requirements; for example, CPMV withstands temperatures of 37 °C for 46 days109 and 60 °C for up to 1 hour110. Thermostability is especially important in areas in which resources for adequate cold chain distribution of vaccines may be limited. Furthermore, plant viruses, owing to their intrinsic stability, can withstand the processing and manufacturing steps of vaccine packaging into new administration technologies, such as microneedles96. VNPs do not only act as carriers, but also as adjuvants, reducing the number of vaccine components. Moreover, the simplicity of VNP-based vaccines can improve health-care worker compliance and reduce the need for training in vaccine administration.

Administration of vaccines with hypodermic needles, which is currently the most frequently used administration technology, has several drawbacks. For example, blood-borne pathogens can easily be spread by needle reuse or mishandling111. Moreover, patients may be apprehensive about needles112. Vaccines delivered by hypodermic needles also rely on liquid formulations for injections, which require a cold chain113.

To overcome these obstacles, vaccines can be encapsulated in polymeric materials. Proteins in the solid state embedded within a polymer matrix further improve the thermal stability of the vaccine, thereby decreasing strict adherence to the cold chain114. Polymeric materials could be used in new vaccine delivery platforms that could supplant hypodermic needles, improving patient compliance and reducing the unintended spread of blood-borne pathogens. Next-generation polymeric delivery platforms that are safe, effective and self-administered could thus greatly strengthen concerted vaccination efforts.

### Polymeric materials

Considerable research has been devoted to the development of biocompatible and biodegradable polymeric materials for vaccines115 (Table 3). These materials can be of synthetic or natural origin, and processing and manufacturing requirements can differ for each polymer, which may restrict the use of biological components. For example, inactivated or live attenuated vaccines can denature at high temperatures, which restricts the processing conditions (some processes require temperatures above 70 °C) required to encapsulate these vaccines in polymeric materials113. By contrast, VLPs and VNPs are chemo- and thermostable, enabling a broader array of device manufacturing conditions116. Fabricated by plant molecular farming, VLPs and VNPs can be produced at large scales for material integration at lower cost, as compared with conventional inactivated or live attenuated vaccines108.

Polymeric materials, developed for next-generation vaccine administration, generally act as delivery vehicles, allowing the controlled release of antigen and adjuvant over weeks to years, depending on the type of material. The antigen release rate can be controlled by tuning the material properties. The presence of degradable bonds in the polymer’s backbone and its hydrophilicity, molecular weight and crystallinity influence the degradation rate of both synthetic and natural polymers. For example, to enable instantaneous burst release, antigens can be encapsulated in rapidly dissolving water-soluble polymers, such as polyvinylpyrrolidone (PVP), which can be processed into microneedles for intradermal vaccine delivery117. A slow and sustained release can be achieved by choosing slowly degrading polymers, such as poly(lactic-co-glycolic acid) (PLGA)118. Increasing the window of antigen delivery can further improve the efficacy of vaccines119,120. By combining burst and slow-release strategies, single-dose regimens can be designed that mimic the multi-dose prime–boost regimen of conventional vaccines121; for example, a vaccine could be encapsulated in both a burst-release polymer and a slow-release polymer, which can then be simultaneously administered.

Delivery of adjuvants is key for vaccines. Some subunit vaccines, such as VLPs or recombinant VLPs and VNPs, which deliver subunit antigens, may not require additional adjuvants because the VLP or VNP itself acts as the adjuvant through recognition and activation of pattern recognition receptors122. However, for moderately or poorly immunogenic antigens, adjuvants need to be added to the formulation. The most frequently used adjuvants are aluminium-based salts (alum)113, which serve both as a delivery vehicle by adsorbing antigen to the surface and as an immunostimulator by inducing pro-inflammatory signalling cascades123. Similarly, polymers such as alginate, hyaluronic acid and PLGA can act as both delivery vehicle and adjuvant owing to their natural inflammatory properties124. Further activation of innate immunity can be achieved through co-delivery of Toll-like receptor (TLR) ligands, such as CpG-oligodeoxynucleotide or monophosphoryl lipid A. Here, materials design is crucial, because a growing body of data suggests that co-delivery of adjuvant and antigen to the same antigen-presenting cell is important to enhance vaccine efficacy and to increase safety by lowering the likelihood of an uncontrolled and unspecific systemic immune response125.

### Vaccine delivery devices and production

As an alternative to subcutaneous injection with hypodermic needles, other routes of administration have been explored, using different device designs and fabrication, which have a considerable effect on immunostimulation. Here, we discuss two delivery constructs for vaccine self-administration: nano- and microparticle formulations, and microneedle patches. Using these devices, vaccines can be self-administered by intradermal, intranasal or oral administration (Fig. 4).

#### Nano- and microparticles

Vaccines can be encapsulated in polymeric nano- and microparticles. For example, polyester-based microparticles, such as PLGA microparticles, can be fabricated starting with an initial water-in-oil dispersion of polymer and antigen, followed by an additional emulsion and solvent removal step, polymer phase separation or spray drying126. By contrast, polyanhydrides require completely anhydrous conditions (solid–oil–oil dispersions) to prevent premature polymer hydrolysis127. Antigens can also be encapsulated in ionic polymers, such as polyphosphazenes and polysaccharides, by crosslinking of the polymers through the addition of oppositely charged multivalent ions in aqueous and ambient conditions. However, these fabrication techniques have several drawbacks, which can limit their use for certain biologics. For example, water-in-oil dispersion has low encapsulation efficiencies, which may limit its use for expensive and difficult-to-produce biologics128. Importantly, the fabrication process can affect antigen stability129; for example, antigens can degrade or unfold owing to interactions at the aqueous–organic interface or to high shear forces during emulsification130.

Nano- and microparticles are often used in self-administered intranasal vaccines. Here, antigens are encapsulated within nano- or microparticles, usually made of poly(lactic acid) or PLGA, and administered as a spray or aerosol, ideally targeting immune cells that reside in the nasal-associated lymphoid tissue of the upper respiratory tract131. However, following intranasal administration, the mucociliary clearance mechanism may clear the deposited vaccine particles, with a half-life of around 20 minutes132. To prevent rapid clearance, the vaccines can be formulated with mucoadhesive polymers, such as chitosan, which can translocate across the nasal mucosal barrier and prolong the residence time to hours or days to improve the immune response133.

#### Microneedles

Microneedles are microscale projections, usually 50 to 900 μm in height, which can pierce the upper layers of the skin and facilitate transdermal delivery of vaccines134. Microneedles do not puncture deep enough to activate the nociceptors within the skin and are therefore considered painless. Microneedles can achieve greater immunological responses, compared with traditional routes of administration, such as intramuscular or subcutaneous injection, owing to the abundant immune cell population in the upper layers of the skin, including the Langerhans cells in the epidermis and the dendritic cells in the dermis135,136. They can also be fabricated to deliver a range of biological vaccines, including inactivated and live attenuated viruses, VLPs and VNPs, protein subunits and DNA137. Importantly, encapsulation in microneeedle patches increases the thermostability of these biologics, which is particularly important for vaccination in regions without proper cold chain infrastructure. Therefore, microneedles are perhaps the most promising device for widespread vaccine distribution and self-administration.

Coated and dissolving microneedles have also been developed. Coated microneedles are designed by coating the surface of microneedles with antigen by dry-coating138 or layer-by-layer electrostatic deposition139. Upon skin puncture, the coated antigen is immediately released to elicit an immunological response. For dry-coating, a stainless steel or silicon microneedle is dip-coated or dry-sprayed with a mixed solution of antigen and carboxymethylcellulose140, polysaccharides141 or synthetic polymers142. A disadvantage of this method is its limited loading capacity134. However, loading can be increased by layer-by-layer electrostatic deposition; here, layers of oppositely charged polymers and proteins or nucleic acids are deposited in an alternating fashion139.

For the fabrication of dissolving microneedles, antigen is encapsulated within the microneedle tips, either in the bulk material or in microparticles. Upon insertion into the upper layers of the skin, the microneedles dissolve at rates dependent on the materials used143. Each material has advantages and disadvantages (Table 3); for example, PVP117 rapidly dissolves upon contact with the interstitial fluid, which decreases the application time of the microneedles, but requires high doses owing to fast clearance rates. Alternatively, chitosan biodegrades slowly, thereby acting as a sustained-release depot. Chitosan microneedles have been shown to induce a more potent and long-lasting immune response compared with conventional intramuscular injection or injection with alum121,144. Dissolving microneedles are usually manufactured by micromoulding145. Here, a mould is first filled with a polymer solution containing the antigen and then placed under vacuum or centrifuged to draw the solution into microcavities. The device is then allowed to dry and harden to form robust microneedle arrays. Water-soluble polymers, such as carboxymethyl cellulose and PVA, manufactured into microneedles under mild conditions, can protect encapsulated antigens from degradation146,147. Alternatively, dissolving microneedles can be fabricated by ultraviolet photo-polymerization of PVP within the mould117 or by high-temperature moulding148; however, these processes can cause antigen degradation, because they require harsher conditions than solvent-based moulding.

Microneedles with plant-produced antigens have already been shown to be effective. For example, a recombinant influenza VLP vaccine (along with a glucopyranosyl lipid adjuvant) administered by the NanoPass MicroJet microneedle, produced equal levels of haemagglutinin inhibition titres and seroprotection in humans, compared with intramuscular injection149. In ferrets, the microneedle formulation led to 100% survival against a H5N1 challenge. Microneedle delivery also enables in situ vaccination; for example, microneedle-delivered CPMV showed greater potency in a mouse model of melanoma compared with intratumoral injection of CPMV, resulting in slower tumour growth and improved overall survival150.

## Outlook

The use of biologics in pharmaceutical applications will continue to grow as expression and purification technologies develop and mature. Plant molecular farming, in particular, is poised to make an impact on an array of applications, and new methodologies, such as the cookie technique, will continue to drive forward its use. Plant molecular farming is currently applied mainly in niche applications; however, recent developments and milestones, such as the production of millions of influenza vaccine doses in record time, has put a spotlight on this manufacturing platform45. Following such successes, it is likely that molecular farming will gain in market share. We have just witnessed the worldwide efforts by laboratories and industry in pivoting their particular platforms and tools towards making vaccines, therapeutics or reagents for the ongoing COVID-19 pandemic or the Zika and Ebola epidemics11,151. Plant molecular farming requires less sophisticated infrastructure than does conventional vaccine production, and thus could be locally deployed in each region, therefore addressing vaccine rollout challenges. Moreover, plant molecular farming could enable the production of vaccines, therapeutics or reagents in space, either on a space station, or during months-long missions to Mars or other extraterrestrial destinations152. The first grants supporting this technology have already been issued, for example, the NASA-funded use of transgenic lettuce and potato leaves to produce growth factors and hormones for astronauts to combat osteoporosis and other spaceflight-related health issues152.

However, expression levels in plant molecular farming must improve to levels similar to those achieved in mammalian or bacterial systems. Expression in bacterial systems was first achieved in the 1970s, ten years before expression in plants153. This gap means that plant molecular farming lags behind other expression systems in terms of expression efficiency, which will be addressed by more efficient plant systems that can produce proteins at larger yields. In addition, downstream processing and purification of biologics need to improve, which remains challenging owing to the complexity of plant cells and tissues (plants produce 30% more solid debris than other expression systems and contain many contaminants, such as phenolics, which interfere with purification). The FDA Critical Path Report states that downstream processes were one of the key reasons that plant-made proteins routinely failed to transition into the clinic or industry154. However, new methods, such as using acqueous two-phase partitioning systems, enzymatic hydrolysis, ultrasound, microwaves, pulsed electric fields, high-voltage electric discharge, ohmic heating, vacuum, subcritical and supercritical fluids, and hydrotropic and deep eutectic solvents are being developed to improve downstream processing of plant material155,156.

Manufacturing of biologics by plant molecular farming is currently performed mainly at small scales by academic research laboratories and mid-sized pharmaceutical companies, at a lower cost than traditional fermentation-based expression systems (up to ten times and a hundred times lower than the production costs in E.coli and CHO cells, respectively)157. Once plant molecular farming reaches agricultural crop scales, the margins are expected to widen even further. The lower costs will help to offset the expensive costs of everyday laboratory research. Thus, molecular farming has the potential to reduce the price of recombinant therapeutics to levels similar to those of everyday over-the-counter medications. Molecular farming could also play a key part in producing biosimilars (biologics that are not identical to the original product, but are equally efficacious) to fabricate cheap pharmaceuticals or biological test reagents for all populations. In addition, decreasing the price of medications would also greatly reduce financial strains on the health-care system. For example, the monoclonal antibodies used in the alleviation of rheumatoid arthritis are currently unaffordable outside health-care systems in developed regions, although the majority of sufferers remain in underdeveloped regions.

Future vaccines should be engineered to be free of cold chain restrictions, for example, by producing orally dosed vaccines and therapeutics in plants. Although the 1990s dream of simply ‘eating your vaccines’ is unattainable (mainly owing to low yield of active ingredient and dose control), transgenic plants consumed by humans, such as potatoes, tomatoes and corn, can be used to not only deliver partially purified and quantified antigen, but also to protect the antigen within a heat-stable environment158. The rigid plant cell wall can protect antigens from degradation within the harsh environment of the gastrointestinal system, while allowing antigen delivery159. The use of orally dosed vaccines can help reduce cold chain requirements and obviate the need for antigen purification and downstream processing62. The amount of land required to vaccinate entire countries by orally dosed vaccines would also be minimal (for example, all of China could be vaccinated with only 40 hectares of land)62.

Cold chain restrictions could also be mitigated by using new delivery agents, such as plant-made VNPs, that are intrinsically thermostable over a range of temperatures and pH levels, allowing their use in microneedle patches (which require high-temperature processing methods110), and in oral vaccines (avoiding denaturation by the acidic gastrointestinal environment160). Finally, new formulation chemistries and next-generation materials are anticipated to enhance formulation stability and improve delivery techniques. For example, the biomineralization of viral vaccines generates a mineral exterior, which improves the thermostability of the viral antigen161. Furthermore, new combinations of excipients can further extend the half-life of vaccines162. Finally, the formulation of biologics in microneedle patches and/or slow-release formulations has tremendous potential for the self-administration of medication.

Thus, platform technologies to produce biologics from living systems continue to develop as new technologies become available, and next-generation nanotechnology and materials science have opened up opportunities in formulation chemistry and the administration of pharmaceuticals.