Article https://doi.org/10.

1038/s41467-022-33516-1

Expression strategies for the efficient

synthesis of antimicrobial peptides
in plastids
Received: 12 April 2022 Matthijs P. Hoelscher 1,2, Joachim Forner 1, Silvia Calderone 1,3,
Carolin Krämer1, Zachary Taylor 1, F. Vanessa Loiacono 1, Shreya Agrawal1,4,
Accepted: 21 September 2022
Daniel Karcher1, Fabio Moratti 1, Xenia Kroop1 & Ralph Bock 1

Check for updates Antimicrobial peptides (AMPs) kill microbes or inhibit their growth and are
1234567890():,;
1234567890():,;

promising next-generation antibiotics. Harnessing their full potential as anti-

microbial agents will require methods for cost-effective large-scale production
and purification. Here, we explore the possibility to exploit the high protein
synthesis capacity of the chloroplast to produce AMPs in plants. Generating a
large series of 29 sets of transplastomic tobacco plants expressing nine dif-
ferent AMPs as fusion proteins, we show that high-level constitutive AMP
expression results in deleterious plant phenotypes. However, by utilizing
inducible expression and fusions to the cleavable carrier protein SUMO, the
cytotoxic effects of AMPs and fused AMPs are alleviated and plants with wild-
type-like phenotypes are obtained. Importantly, purified AMP fusion proteins
display antimicrobial activity independently of proteolytic removal of the
carrier. Our work provides expression strategies for the synthesis of toxic
polypeptides in chloroplasts, and establishes transplastomic plants as efficient
production platform for antimicrobial peptides.

Antimicrobial peptides (AMPs) are peptides that kill microbes or hydrophobic and hydrophilic groups that fold into an amphipathic
inhibit their growth1–3. AMPs represent an important part of defense conformation, where one side of the molecule is hydrophobic and the
strategies against pathogenic microbes, and are produced by both other side hydrophilic1. The positive net charge of AMPs promotes
eukaryotic and prokaryotic life forms, although the majority of iden- electrostatic interactions with negatively charged bacterial mem-
tified AMPs are from animals4. branes, but not with neutral eukaryotic membranes1,8,9. The hydro-
Due to the increasing prevalence of multidrug-resistant bacteria, phobic regions of AMP molecules facilitate binding and integration
AMPs have received considerable interest as an alternative to currently into bacterial membranes1,8. Negatively charged and uncharged AMPs
used antibiotics5,6. In addition to their potential use to treat infectious that kill bacteria also exist, although they are rare. Only 5.7% of bac-
diseases in humans, AMPs are also considered as alternatives to anti- tericidal AMPs are uncharged and only 6% have a charge below zero4
biotics as growth promoters in animal husbandry7. (APD accessed in February 2022). Since purification from the native
AMPs are generally short polypeptides. The average length of all host is usually not feasible and direct chemical synthesis is far
AMPs in the Antimicrobial Peptide Database (APD) is 33.19 amino too expensive (with the possible exception of some very small AMPs
acids4 (database accessed in February 2022). The vast majority that do not require disulfide bridges or other post-translational
of AMPs have a positive net charge and are composed of a mix of modifications), recombinant expression using transgenic methods

1
Max-Planck-Institut für Molekulare Pflanzenphysiologie, Am Mühlenberg 1, D-14476 Potsdam-Golm, Germany. 2Present address: Utrecht University, Phar-
maceutical sciences, Pharmaceutics, Universiteitsweg 99, 3584 CG Utrecht, Netherlands. 3Present address: Centre for Research in Agricultural Genomics
(CRAG), CSIC-IRTA-UAB-UB, Campus UAB, Bellaterra, 08193 Barcelona, Spain. 4Present address: Neoplants, 630 Rue Noetzlin Bâtiment, 91190 Gif-sur-
Yvette, France. e-mail: [email protected]

Nature Communications | (2022)13:5856 1

Article https://doi.org/10.1038/s41467-022-33516-1

represents the most promising strategy for commercial-level AMP only 29 amino acids39, being the smallest example. However, all of
production10. these small proteins are subunits of protein complexes (photo-
In spite of their bactericidal activity, recombinant production of synthetic complexes residing in the thylakoid membrane, chloroplast
some AMPs in bacteria is possible11. However, especially for the large- ribosomes), and the rapid (often co-translational) incorporation of
scale and cost-effective commercial production, plants represent the these small proteins into large multiprotein complexes likely protects
most promising production platform. Transgenic plants have been them from proteolytic degradation.
shown to facilitate the synthesis of large quantities of biopharmaceu- In this work, we systematically compare multiple approaches to
ticals at low costs, with the option to scale up production by simply maximize AMP expression levels in transplastomic tobacco plants, and
expanding the planted area12,13. Both stably transformed and tran- simultaneously, reduce the toxic effects associated with high-level
siently transformed plants can be used for recombinant protein pro- AMP accumulation in chloroplasts. Based on the above considerations,
duction. Transient expression offers the benefit that negative effects of we hypothesize that embedding AMPs into larger polypeptides will
transgene expression on growth can be avoided, and the response increase their stability, because they will not be recognized as PreP
time for development of new therapeutics is short13. Stable expression and/or OPP substrates. In order to increase the size of AMPs, we pursue
systems rely on genetic transformation of either the nuclear or the two parallel strategies: (i) the fusion of multiple antimicrobial peptides
plastid genome. For production purposes, they offer a number of connected by flexible linkers, and (ii) the use of the small ubiquitin-like
significant advantages: The harvested biomass is free of transgenic modifier (SUMO) as a cleavable fusion partner.
pathogenic microbes (Agrobacterium or viruses), there is no require-
ment for transformation of every new batch of plant material (likely Results
also resulting in much greater batch-to-batch consistency in protein Design of AMP fusion constructs for transplastomic expression
accumulation levels) and substantially reducing cost14, and facilitates Given the above-described concerns related to AMP toxicity and
scale-up to large greenhouse or open field cultivation15. The use of instability due to their small size, we sought to test the idea that fusion
specialized promoter systems can trigger expression in all tissues, proteins of several AMPs can be expressed to high levels in chlor-
restrict transgene expression to specific tissues or enable inducible oplasts. In addition, combination of different AMPs could result in
expression16–21. Expression from the plastid (chloroplast) genome is a synergistic antimicrobial effects40–42, thus providing more potent anti-
particularly attractive method of heterologous protein production in microbials. In bacteria, recombinant production of AMPs is frequently
plants, because chloroplasts can accumulate very large amounts of performed as fusions to carrier proteins like glutathione S-transferase
foreign protein22–24, and offer the additional benefits of transgene (GST), thioredoxin or small ubiquitin-like modifier (SUMO) – a strategy
exclusion from pollen transmission and the absence of epigenetic that increases AMP stability and solubility, and reduces toxicity to the
transgene silencing25. Production of AMPs or antimicrobial proteins in host10,33. The carrier proteins can be removed after purification33,43, and
plants has been attempted by transient expression, stable nuclear production of AMPs as SUMO fusions and subsequent cleavage and
transformation and plastid transformation26–32. purification of the peptide were demonstrated in E. coli44. SUMO fusion
When considering possibilities to develop strategies for the pro- and subsequent removal was also used to improve the production of
duction of AMPs in plants by expression from the plastid genome (i.e., the antiviral protein cyanovirin in E. coli45. Fusion proteins comprising
in transplastomic plants), several important considerations come into multiple AMPs have been described in bacteria33,41,42,46,47, but produc-
play. In view of the prokaryotic origin of plastids, a serious concern is tion in chloroplasts has not been attempted yet.
that antimicrobial peptides may damage the plastid (and its mem- Nine AMPs were selected as promising candidates for the con-
branes) in a similar manner as they kill bacteria. Problems with AMP struction of fusion proteins and expression in tobacco chloroplasts
production in bacteria have been extensively studied, and different (Supplementary Data 1). The AMPs cgMolluscidin, CXCL9 and ubiqui-
solutions have been developed to enhance accumulation and reduce cidin were selected because they possess a very strong positive charge,
toxicity10,33. Recombinant AMP production in bacteria is commonly a property that could make them more selective towards negatively
attempted by using an inducible expression system, to allow unhin- charged bacterial membranes48–52 (Supplementary Data 1). They are
dered growth of the bacterial culture before induction of AMP also relatively large with 55, 103 and 59 amino acids, respectively
synthesis10,33. In fact, inducible expression was key to the heterologous (Supplementary Data 1). Six other AMPs, novispirin G10, esculentin‐1‐
production of most of the AMPs that could be expressed in bacteria, OA1, andersonin‐Y1, wallaby antimicrobial 1 (WAM1), dermaseptin S4
indicating that the inherent toxicity of AMPs to prokaryotic produc- (derivative K4K20-S4) and polyphemusin I were selected, because their
tion hosts represents a serious obstacle33,34. minimal inhibitory concentrations (MIC) against both gram-positive
Another significant concern is that AMPs might be unstable inside and gram-negative bacteria were among the lowest to be found in the
plastids due to their small size, the presence of hydrophobic domains literature4,53–58 (Supplementary Data 1).
(which can be problematic in plastid transgene expression35). AMPs ince plastids harbor proteases that specifically recognize and
share these properties with the small plastid targeting peptides that degrade small peptides between 10 and 65 amino acids36,37, we antici-
are rapidly degraded after protein import has been completed5,36,37. pated that single AMPs would be prone to degradation. We, therefore,
In fact, AMPs and transit peptides for protein import into plastids prepared single AMP expression constructs only for three of the
and mitochondria are so similar that AMPs are considered as possible selected AMPs: CXCL9 (with 103 amino acids the largest AMP included
evolutionary ancestors of organellar targeting peptides38. The pre- in our study and the only one above the 65 amino acid threshold;
sequence protease (PreP) in plastids and mitochondria efficiently Supplementary Data 1), and HA epitope-tagged versions of novispirin
degrades cleaved-off transit peptides with a size of 10-65 amino acids37, and WAM1 (with 34 and 52 amino acids, respectively; Fig. 1a). In
suggesting that polypeptides of similar size that are recombinantly addition, to overcome the size-selective degradation, we designed a
expressed in plastids will also be degraded by PreP. PreP displays a set of constructs for the expression of enlarged polypeptides by fusing
preference for positively charged residues at the cleavage site36, which multiple AMPs, separated by flexible linkers of 5 or 15 amino acids
could further contribute to degradation of positively charged (catio- (Fig. 1a; Supplementary Table 1 and Supplementary Data 2). The use of
nic) AMPs. In addition, plastids harbor an oligopeptidase (organellar flexible linkers can preserve the functionality of fused proteins, by
oligopeptidase, OOP) that degrades peptides of 8-23 amino acids36 and allowing independent movement59. In this way, the fusion of different
potentially also contributes to the instability of small polypeptides AMPs also offers the potential to yield synthetic antimicrobial poly-
expressed in plastids. Notably, the chloroplast genome encodes a few peptides that are more potent than natural AMPs by synergistically
small proteins, with PetN, a subunit of the cytochrome b6f complex of combining different modes of action (Supplementary Data 1).

Nature Communications | (2022)13:5856 2

Article https://doi.org/10.1038/s41467-022-33516-1

Fig. 1 | Constitutive AMP and fAMP expression from the tobacco plastid gen- spacer between the trnfM and trnG genes in the tobacco plastid genome. Restric-
ome. a Sequences encoding AMPs or fAMPs were placed under the control of the tion sites (BglII) and the binding site of the probe for RFLP analysis to verify the
Nicotiana tabacum plastid ribosomal RNA operon promoter (NtPrrn), the T7 phage transplastomic state are indicated. For information on AMPs and fAMPs, see Sup-
gene10 leader (T7g10L) as 5′ UTR and the 3′ UTR of the atpA gene from Chlamy- plementary Tables 1, 2 and Supplementary Data 1 and 2. b Seed assays to confirm
domonas reinhardtii (CrTatpA). The spectinomycin resistance gene aadA for homoplasmy of transplastomic AMP-expressing lines. Seeds were sown on medium
selection of transplastomic lines was driven by the Chlamydomonas psaA promoter with (Spec+) or without (Spec−) spectinomycin, and photos were taken after two
and 5′ UTR (CrPpsaA), and the 3′ UTR of the Chlamydomonas atpB gene (CrTatpB). weeks. In the case of 3‐Molluxubin‐5 (3‐M‐5), 3‐Molluxubin‐15 (3‐M‐15) and 3‐
Different combinations of AMPs were linked with linkers of 5 or 15 amino acids Novescandin‐5 (3‐N‐5), seeds were obtained from grafted plants. 3‐N‐15: 3‐Noves-
(pink). The constructs were flanked by homology regions for integration into the candin‐15.

Expression constructs were generated in which combinations of only be achieved by additionally varying the encoded amino acid
three, six or nine different AMPs were fused with linkers of 5 or 15 sequences (Supplementary Table 1).
amino acids (Fig. 1a; Supplementary Data 2). The resulting repeated To facilitate protein detection and purification, all fusions were
use of linkers necessitated the systematic introduction of sequence equipped with N-terminal HA and C-terminal 6xHis tags. The combi-
variation at the DNA level (see Methods; Supplementary Table 1) to nations of three different antimicrobial peptides will subsequently be
prevent construct instability in the chloroplast genome by homo- referred to as 3-Molluxubin (3-M), 3-Novescandin (3-N) and
logous recombination on directly repeated identical sequences60,61. 3-Wamdeposin (3-W). The names are portmanteaus of the AMPs they
For the 15 amino acid linkers, sufficient DNA sequence variation could are composed of (Supplementary Data 2), followed by the number 5 or

Nature Communications | (2022)13:5856 3

Article https://doi.org/10.1038/s41467-022-33516-1

15 to distinguish the 5 and 15 amino acid linker versions (e.g., 3-Wam- delays in growth and development (Fig. 2a). In some cases, the linker
deposin-5). The fusion of 6 AMPs was named 6-MW (combining 3-M length influenced the plant phenotype, albeit not in a consistent
and 3-W) and the fusion of all 9 AMPs was called 9-MNW (combining 3- manner. For 3-Molluxubin, the use of 15 amino acid linkers resulted in
M, 3-N and 3-W; Supplementary Data 2). paler plants compared to those obtained by using the 5 amino acid
linkers. By contrast, for 3-Novescandin, the 15 amino acid linkers per-
Constitutive transplastomic expression of AMPs and AMP mitted growth on soil, whereas the 5 amino acid linkers did not
fusions (Fig. 2a). 3-Novescandin-15 plants grew very slowly, producing seeds
In an attempt to achieve maximal transgene expression levels, we five to nine months after their transfer from in vitro culture to soil.
initially utilized the strongest gene expression elements known for Plants expressing 3-Molluxubin-5, 3-Molluxubin-15 and 3-Novescandin-
chloroplasts: the tobacco (Nicotiana tabacum) plastid ribosomal RNA 5 could grow on soil only when grafted onto a wild-type stock that
operon promoter (Prrn) in combination with the gene10 leader from provided nutrients to the mutant shoots, a strategy that ultimately
the coliphage T722,62 (Fig. 1a). The DNA sequences encoding the AMPs permitted seed production. Transplastomic plants expressing
were codon optimized to match the preferred codon usage of tobacco 3-Wamdeposin or any of the larger fusions (6-MW and 9-MNW) were
chloroplasts63. white and unable to grow on soil, irrespective of the linker length
The constructs were introduced into the tobacco plastid genome (Fig. 2a), thus precluding the production of seeds.
by particle bombardment (biolistic transformation) with DNA con- The production of seeds from several lines enabled the additional
structs containing the AMP transgenes inserted into a standard vector assessment of the homoplasmic state of the transplastomic plants, by
for chloroplast transformation64 (Fig. 1a). The vector (pDK323) con- conducting inheritance assays67. When the transplastomic seeds were
tains a chimeric spectinomycin resistance gene (aadA) as selectable sown on spectinomycin-containing medium (Fig. 1b), no phenotypic
marker for isolation of transplastomic lines65. Regeneration of trans- segregation was observed, suggesting that parent plants and T1 off-
plastomic lines on spectinomycin-containing medium resulted in the spring were homoplasmic.
formation of plantlets that showed widely different pigmentation
phenotypes. Green regenerants were obtained from some constructs, Expression of AMP and fAMP transgenes
while the transplastomic lines obtained with other constructs dis- Transgene expression at the mRNA level was assessed by northern blot
played various degrees of pigment deficiencies, ranging from pale analysis. All transgenes were expressed and accumulated mRNAs
green to completely white (Fig. 1b; Fig. 2a, b). matching the expected sizes (Fig. 2c), in addition to some larger
The constructs expressing WAM1 and novispirin produced green transcript species that likely originated from read-through transcrip-
plantlets. Leaves from these plantlets were subjected to another round tion (a phenomenon that is pervasive in plastids and has been
of regeneration, and DNA was isolated from the resulting plants for observed with transgenes inserted into the same genomic region in
restriction fragment length polymorphism (RFLP) analyses to confirm previous studies68,69).
successful transformation of the chloroplast genome and assess the Protein accumulation of AMPs and fAMPs (hereafter collectively
attainment of the homoplasmic state (i.e., the presence of a homo- referred to as (f)AMPs) was examined by immunoblotting using an
geneous population of transformed genomes and the absence of resi- antibody against the HA-tag. While no anti-HA signal could be
dual wild-type copies of the highly polyploid plastid genome66). Several observed for constructs expressing the single AMP novispirin as well
lines were identified for which the RFLP analysis showed a strong band as untagged CXCL9 or the large constructs combining 6 or 9 dif-
with the expected size suggesting correct integration of the transgene ferent AMPs, (Fig. 2d), all other transplastomic lines showed bands
by homologous recombination (Supplementary Fig. 1). corresponding to the expected size of the respective AMPs and
Regenerated plantlets from transplastomic lines expressing 3- fAMPs, and additional weaker signals corresponding in size to pos-
Molluxubin, 3-Novescandin and CXCL9 initially had variegated leaves sible multimers (Fig. 2d). A notable exception were WAM1 trans-
with dark green, light green and/or white sectors, and RFLP analysis of plastomic plants that showed a strong signal at twice the expected
this material revealed the simultaneous presence of hybridization size, possibly suggesting an unusually strong tendency to dimerize
signals for the transformed and the wild-type plastid genomes (het- even under the denaturing conditions of purification and SDS-PAGE
eroplasmy; Supplementary Fig. 1). After several additional regenera- electrophoresis.
tion cycles, individual regenerants showed apical formation of
uniformly pale shoots and leaves, and RFLP analysis of DNA extracted Control of fAMP expression using the RNA amplification-
from these phenotypically homogeneous tissues resulted in a single enhanced riboswitch system
band corresponding in size to the transformed plastid genome (Sup- The observed phenotypes of plants constitutively expressing fused or
plementary Fig. 1). When the uniformly pale shoots were rooted, the large single AMPs suggested strong deleterious effects of AMP accu-
pale phenotype remained stable over multiple propagation cycles, mulation on plant growth and chloroplast development. To overcome
indicating that the lines had been purified to homoplasmy (Fig. 2a, b). this problem, inducible expression was attempted to allow for proper
Transplastomic plants expressing 3-Wamdeposin-5, 3-Wamdeposin-15, growth and assembly of the photosynthetic apparatus prior to fAMP
6-MW-5, 6-MW-15, 9-MNW-5 and 9-MNW-15 resulted in completely accumulation. The chosen inducible RNA amplification-enhanced
white plants that ultimately also could reach homoplasmy (Supple- riboswitch (RAmpER) system17 is entirely chloroplast based and con-
mentary Fig. 1; Fig. 2a). trols transgene expression by a theophylline-inducible riboswitch70.
As expected, white transplastomic plants were incapable of pho- Upon induction (i.e., watering of plants with theophylline solution),
toautotrophic growth when transferred from synthetic sugar- translational activation leads to the synthesis of the bacteriophage T7
containing medium to soil and thus died quickly. This was also the RNA polymerase, which subsequently transcribes the gene of interest
case with several transplastomic lines that were very pale green such as that is controlled by the T7 promoter17 (PT7).
the plants expressing the large single AMP CXCL9 (Fig. 2b). By con- To eliminate the need to assemble large transformation vectors
trast, plants expressing the single small HA-tagged AMPs WAM1 or containing all elements of the RAmpER system (including the difficult-
novispirin were green and displayed only a slight delay in growth to-clone T7 RNA polymerase gene), we first generated a transplastomic
compared to wild-type plants (Fig. 2b). line that harbors the riboswitch-controlled T7 RNA polymerase
Transplastomic plants with transgenes encoding fused AMPs (RT7RP) in the chloroplast genome (Fig. 3b). This line (Nt-DK320) was
(fAMPs) showed strong phenotypes, ranging from light green to used as recipient line for subsequent supertransformation experi-
completely white, and those that survived in soil exhibited extreme ments with constructs encoding the fAMP constructs under control of

Nature Communications | (2022)13:5856 4

Article https://doi.org/10.1038/s41467-022-33516-1

Fig. 2 | Phenotypes of transplastomic plants that constitutively express AMPs plants two months after sowing, see Supplementary Fig. 4. c Northern blot analysis
and fAMPs and analysis of transgene expression. a Plant phenotypes of trans- to compare mRNA accumulation for AMP and fAMP transgenes in transplastomic
plastomic lines expressing fAMP combinations of 3, 6 or 9 different AMPs con- tobacco lines (see Methods section). MB: methylene blue staining as loading con-
nected by 5 or 15 amino acid linkers (length of linker is indicated on the left of the trol. The blot was performed twice with similar results. d Western blot analysis of
pictures; cf. Supplementary Table 1, Supplementary Data 1 and 2). Plants incapable AMP accumulation (see Methods section). Samples of total protein extracts (20 µg)
of photoautotrophic growth on soil are shown in tissue culture boxes. Photos were were loaded, and fAMPs were immunodetected with an anti-HA antibody. Note that
taken three months, or 4.5 months (6-MW-15) or six months (3-W-5) after transfer of CXCL9 is not tagged. The Coomassie-stained membrane is shown below the
(apical) shoot cuttings to fresh medium. 3‐N‐5 was the only fAMP line capable of immunoblot as a loading control. Three independent blots were performed with
growing on soil. The photo was taken five months after transfer from tissue culture. similar results. Plant material for northern and western blots was produced in
b Phenotypes of transplastomic plants expressing single AMPs. Shown are (from sterile culture to enable side-by-side comparison to lines that do not survive on soil.
left to right) a CXCL9 plant three months after transfer of a shoot into a new box C: CXCL9, N: HA-Novispirin, W: HA-WAM1, 3M: 3‐Molluxubin, 3W: 3‐Wamdeposin,
with synthetic medium, HA‐WAM1, HA‐Novispirin and a wild‐type plant one month 3N: 3‐Novescandin, 6MW: 3M + 3W, 9MNW: 3M + 3N + 3W. Source data are pro-
after sowing on soil, and a wild‐type plant after one month of growth on synthetic vided as a Source Data file.
medium. Scale bars: 2.5 cm. For images of HA-Novispirin, HA-WAM1 and wild-type

the PT7 (Fig. 3a, b). The resulting supertransformed plants were named in the absence of the T7 RNA polymerase, two control lines were
RAmpER::fAMP lines. generated. To this end, two of the constructs used for super-
Some leakiness (i.e., basal expression in the non-induced state) transformation of the RT7RP recipient line, were also used to trans-
has previously been reported for the RAmpER system17,71. To determine form wild-type plants. The resulting transplastomic lines were named
if this leakiness is caused by T7 RNA polymerase accumulation in the PT7::fAMP, as opposed to the RAmpER::fAMP lines containing the
non-induced state or rather by transgene transcription from PT7 even RT7RP (Fig. 3).

Nature Communications | (2022)13:5856 5

Article https://doi.org/10.1038/s41467-022-33516-1

Fig. 3 | Application of the RAmpER system to control expression of fAMPs from rbcL 3′ UTR (CrTrbcL). The T7 RNA polymerase transgene is under the control of the
the tobacco plastid genome. a Construct design (cf. Fig. 1a). The fAMP-encoding C. reinhardtii psbA promoter (CrPpsbA), the theophylline-responsive riboswitch and
sequences were inserted between the T7 promoter (PT7) with the T7 gene10 leader the 3′ UTR from the tobacco rps16 gene (NtTrps16). Supertransformation of
(g10L) and the C. reinhardtii atpA 3′ UTR (CrTatpA) followed by the T7 gene10 3′ transplastomic plants harboring the T7 RNA polymerase cassette with PT7-driven
UTR (TT7). For information on the fAMP combinations and the linkers (of 5 or 15 fAMP transgenes was performed to create transplastomic lines with RAmpER-
amino acids; pink), see Supplementary Table 1 and Supplementary Data 2. Note that controlled fAMP expression. c Seed tests of transplastomic lines expressing fAMPs
the constructs driven by PT7 have the opposite orientation in the plastid genome under RAmpER or PT7 control. Surface sterilized seeds were sown on medium with
compared to the constitutive constructs (Fig. 1), to reduce potential read-through or without spectinomycin (Spec). Photos were taken after two weeks, the wild-type
transcription from upstream plastid genes. b Physical map of the region in the controls are also used in Fig. 1b. RAmpER::3-W-5 plants were incapable of auto-
plastid genome of the transplastomic recipient line Nt-DK320 that harbors the trophic growth on soil, and therefore, seeds were obtained by grafting a shoot
riboswitch-controlled T7 RNA polymerase gene targeted to the intergenic spacer raised in sterile culture onto a wild-type stock. 3-M-15: 3-Molluxubin-15, 3-N-5: 3-
between ycf3 and psaA by selection for kanamycin resistance provided by a chi- Novescandin-5, 3-W-5: 3-Wamdeposin-5, 6-MW-5: 3-M-5 + 3-W-5.
meric aphA-6 gene controlled by the C. reinhardtii psaA promoter (CrPpsaA) and

The homoplasmic state for the transplastomic RAmpER::fAMP wild-type-like pigmentation and growth rate (Fig. 4a). Replacement of
and PT7::fAMP plants (Fig. 3a) was confirmed by RFLP analysis (Sup- the constitutive expression signals by RAmpER effectively shifted the
plementary Fig. 2) and seed test (Fig. 3c) as described before. plant phenotypes from pale green to darker green or from white to
pale green (Fig. 4a). RAmpER::3-Molluxubin-15 and RAmpER::3-
Phenotypes of RAmpER::fAMP plants Novescandin-5 were green and capable of growing autotrophically on
Transplastomic plants in which production of fAMPs was controlled by soil, albeit at slower rate than wild-type plants. While Prrn::3-Wamde-
the RAmpER system showed markedly improved growth phenotypes posin-5 and Prrn::9-MNW-15 plants were white, RAmpER::3-Wamde-
in the non-induced state compared to strong constitutive posin-5 and RAmpER::9-MNW-15 plants were pale green, but still
fAMP expression, in all cases. However, they still did not display unable to grow on soil. A graft of RAmpER::3-Wamdeposin-5 onto a

Nature Communications | (2022)13:5856 6

Article https://doi.org/10.1038/s41467-022-33516-1

Fig. 4 | Comparison of transplastomic plants expressing fAMP constructs (P), RAmpER (uninduced) (R) and PT7 (T) plants (see Methods). MB: methylene blue
controlled by different promoter systems. a Phenotypes of transplastomic staining (loading control). This blot was performed twice with similar results.
plants. Two genotypes were used as recipients for introduction of fAMP constructs: d Western blot analysis of fAMP accumulation. Strong constitutive expression from
the wild type and a transplastomic line harboring the riboswitch-controlled T7 RNA Prrn (P) is compared with basal RAmpER expression (in the non-induced state; R)
polymerase (RT7RP). The promoters used to drive fAMP expression is given in each and low-level background expression from PT7 (in the absence of the T7 RNA
row. Shown are transplastomic plants expressing fAMPs under the control of the polymerase; T). Total protein samples (20 µg) were loaded, and fAMPs were
Prrn promoter (images from Fig. 2a), transplastomic plants expressing fAMPs immunodetected with an anti-HA antibody. The Coomassie-stained membrane is
under RAmpER control (uninduced), and two PT7-controlled fAMP lines in the wild- shown below the immunoblot as loading control. Note that pale and white lines
type background. Photos of soil-grown plants were taken one month after sowing. have reduced amounts of the large subunit of RuBisCO (at ∼55 kDa). Two inde-
Scale bars: 5 cm. 3-M: 3-Molluxubin, 3-N: 3-Novescandin, 3-W: 3-Wamdeposin, 6- pendent blots were performed with similar results. Plant material for northern and
MW: 3-M + 3-W, 9-MNW: 3-M + 3-N + 3 W. The numbers 5 or 15 indicate the lengths western blots was produced in sterile culture to enable inclusion of lines that do not
of the amino acid linkers separating the AMPs (see Supplementary Table 1 and survive on soil. 3M15: 3-Molluxubin-15, 3N5: 3-Novescandin-5, 3W5: 3-Wamdeposin-
Supplementary Data 2). b Images of the plants in (a) two months after sowing. Scale 5, 6MW5: 3M5 + 3W5, 9MNW15: 3M15 + 3N15 + 3W15. Source data are provided as a
bars: 5 cm. c Northern blot analysis comparing the fAMP expression levels in Prrn Source Data file.

wild-type stock survived on soil, produced seeds, and could therefore When grown side by side, from seeds, PT7::3-Novescandin-5
be included in the seed assays (Fig. 3c). While Prrn::6-MW-5 plants were plants were indistinguishable from wild-type plants, while RAmpER::3-
white, RAmpER::6-MW-5 plants were green, grew on soil and produced Novescandin-5 plants showed a delayed growth. PT7::3-Wamdeposin-5
seeds, although growth was severely retarded (Fig. 4). plants grew on soil and were green, unlike the RAmpER::3-

Nature Communications | (2022)13:5856 7

Article https://doi.org/10.1038/s41467-022-33516-1

Wamdeposin-5 lines (Fig. 4). In contrast to PT7::3-Novescandin-5, the SUMO-AMP plants, like exhaustion of the protein synthesis
PT7::3-Wamdeposin-5 showed a clear delay in growth (Fig. 4). capacity22.

fAMP accumulation in RAmpER and PT7 plants Accumulation of SUMO fusion proteins
Northern and western blot analyses were performed to compare the The improved growth of RAmpER::SUMO-3-Wamdeposin-5 compared
levels of fAMP mRNA and protein accumulation in non-induced to RAmpER::3-Wamdeposin-5 plants could be due to reduced fAMP
RAmpER::fAMP and PT7::fAMP plants with those in the corresponding accumulation upon fusion to SUMO. However, immunoblot analysis
constitutively expressing Prrn::fAMP plants. As expected and in revealed the opposite, as a much stronger signal was seen for RAm-
agreement with the phenotypes of the lines (Fig. 4a), fAMP mRNA and pER::SUMO-3-Wamdeposin-5 than for RAmpER::3-Wamdeposin-5
protein levels were the highest in Prrn::fAMP plants, substantially lower (Fig. 6c). For all fAMPs tested, the SUMO fusion proteins accumulated
in non-induced RAmpER::fAMP plants and undetectable in PT7::fAMP to higher levels than the unfused versions (Fig. 6c).
plants (Fig. 4c, d). No consistent difference was observed in the steady-state abun-
Initial extensive attempts to purify fAMPs from RAmpER::fAMP dance of the transgenic transcripts between the versions with and
plants were unsuccessful. We found that solubilization of the fAMPs without SUMO, suggesting that the observed increase in protein
from plant tissues required the use of detergents, and subsequent accumulation is caused by improved stability of SUMO fusion proteins
separation of proteins and detergent was unsuccessful. The presence (Fig. 6b, c).
of detergents in the protein extracts is highly undesirable, because
they themselves have antimicrobial activity and thus would interfere Purification of SUMO fusion proteins
with any downstream test for biological activity. Protein purification was performed successfully for SUMO-HA-WAM1,
SUMO-HA-Novispirin and SUMO-GFP using buffers without deter-
SUMO fusions for enhanced solubility and reduced toxicity gents. All three proteins were detected in the soluble phase, indicating
of fAMPs that fusion to SUMO improves solubility of the AMPs, thus obviating
To improve fAMP solubility while potentially reducing their toxicity the need to use detergents (Fig. 7a). In Prrn::SUMO-GFP and RAm-
towards the host, fusions of the fAMPs to the small ubiquitin‐like pER::SUMO-GFP plants, SUMO-GFP was readily detectable by Coo-
modifier (SUMO) were constructed. To this end, vectors were pro- massie staining (Fig. 6c) and accumulated to similar levels as the most
duced for C-terminal fusion of the protein of interest to N-terminally abundant endogenous protein, the large subunit of RuBisCO.
6xHis-tagged SUMO (12.8 kDa, 111 amino acids) under the control of Batch purification with Ni-NTA agarose yielded full-length
the T7 or Prrn promoter (Fig. 5a, b). SUMO-GFP in the eluate (Fig. 7a). This was also the case with
Seven SUMO fusion constructs with different expression elements SUMO-HA-WAM1 and SUMO-HA-Novispirin, although an additional
were generated (Fig. 5a, b). Constructs with PT7 were introduced in band corresponding in size to free SUMO was present in the eluate
both the RT7RP recipient line to generate RAmpER::SUMO-(f)AMP from the SUMO-HA-WAM1 purification experiments (Fig. 7a). Large-
plants and wild-type plants. Transplastomic plants were obtained, and scale purification followed by quantification of the purified protein
their homoplasmic state was confirmed by RFLP analysis and seed test products revealed yields of approximately 56 µg per g leaf material
as described previously (Supplementary Fig. 3 and Fig. 5c). for Prrn::SUMO-HA-WAM1, 54 µg per g leaf material for Prrn::SUMO-
HA-Novispirin, 41 µg per g leaf material for RAmpER::SUMO-HA-
Phenotypes of transplastomic plants expressing SUMO-(f)AMPs WAM1 (non-induced), and 654 µg per g leaf material for Prrn::SUMO-
Fusing 3-Wamdeposin-5 to SUMO, in conjunction with reduced GFP. Based on ratios between the molecular weights of 6xHis::SUMO
expression levels, reduced the toxicity of this polypeptide to the plant. (12,794 Da) and the AMPs or protein fused to it (HA-WAM1: 5,864 Da,
RAmpER::SUMO-3-Wamdeposin-5 plants were visibly darker green HA-Novispirin: 3,794 Da, GFP: 27,056 Da), we also calculated the
than plants expressing the same construct without the SUMO tag yields without the SUMO part of the recombinant proteins. For
(Fig. 6a). Similarly, PT7::SUMO-3-Wamdeposin-5 plants were con- Prrn::SUMO-HA-WAM1, yields were approximately 18 µg of HA-WAM1
siderably less retarded in growth than the PT7::3-Wamdeposin-5 plants per gram leaf material, for RAmpER::SUMO-HA-WAM1 (non-induced),
(Fig. 6a). By contrast, Prrn::SUMO-3-Wamdeposin-5 plants were com- yields were approximately 13 µg HA-WAM1 per gram leaf material, for
pletely white, resembling the phenotype of Prrn::3-Wamdeposin-5 Prrn::SUMO-HA-Novispirin, yields were approximately 12 µg HA-
plants. Novispirin per gram leaf material, and for Prrn::SUMO-GFP, yields
Although RAmpER::SUMO-3-Wamdeposin-5 plants were unable to were approximately 444 µg GFP per gram leaf material. Quantifica-
grow autotrophically, seeds could be obtained from a heteroplasmic tions are provided in Supplementary Data 3.
plant that developed a shoot exhibiting the pale color typical of this Next, we tested whether the proteins of interest could be released
construct upon segregation into homoplasmy. by cleavage of the purified fusion proteins with recombinant SUMO
For single AMPs, contrasting effects of SUMO fusion were protease, a highly specific enzyme which hydrolyzes the peptide bond
observed. Whereas growth of Prrn::HA-WAM1 plants was only mildly at the C-terminus of the SUMO domain. After incubation with the
retarded, Prrn::SUMO-HA-WAM1 plants grew slowly and were pale protease, the expected size shift was observed that corresponded to
green (Fig. 6). However, when the latter construct was controlled by conversion of full-length SUMO-tagged protein to free SUMO (Fig. 7b).
RAmpER, this phenotype was much alleviated and only consisted of a The efficiency of digestion based on image quantification is approxi-
slight growth delay (Fig. 6a). RAmpER::SUMO-HA-WAM1 plants in the mately 80% for SUMO-GFP and SUMO-Novispirin, and 60% for SUMO-
T0 and T1 generations were male sterile (4 independently generated HA-WAM1. Quantifications are provided in Supplementary Data 3. For
transplastomic lines) and had to be pollinated with wild-type pollen to digested SUMO-GFP, free GFP was readily detected on the Coomassie-
obtain seeds. stained western blot membrane. No band was seen for HA-WAM1 by
While Prrn::HA-Novispirin plants were nearly indistinguishable Coomassie staining, but the peptide was observed by immunodetec-
from wild-type plants, Prrn::SUMO-HA-Novispirin plants grew slower tion (Fig. 7b). After cleaving SUMO-HA-Novispirin a low-molecular
and new leaves were pale green (Fig. 6a), presumably because SUMO weight band was detected on the Coomassie stained western blot
fusion caused increased AMP accumulation. RAmpER::SUMO-GFP membrane, which migrated slightly slower than the predicted size of
plants displayed wild-type-like growth, while Prrn::SUMO-GFP plants HA-Novispirin. However, this band was recognized only weakly by the
were delayed in growth, presumably due to other mechanisms than in anti-HA antibody (Fig. 7b).

Nature Communications | (2022)13:5856 8

Article https://doi.org/10.1038/s41467-022-33516-1

Fig. 5 | Transplastomic expression of AMPs and fAMPs as SUMO fusions. orientation in the plastid genome in comparison to the Prrn-driven transgenes,
a Physical maps of constructs for the expression of 6xHis-SUMO fusions under to reduce potential read-through from upstream plastid genes. c Seed tests
the control of the T7 RNA polymerase promoter (PT7). For information on con- to confirm homoplasmy of transplastomic plants expressing SUMO fusions.
struct design and the expression elements used, see Figs. 1a and 3a. b Physical Because RAmpER::SUMO-3-Wamdeposin-5 plants were unable to survive on soil,
maps of constructs for the expression of 6xHis-SUMO fusions under the control seeds were obtained from a heteroplasmic plant that segregated into dark green
of the plastid rRNA operon promoter (Prrn) and the gene10 leader (T7g10L) from (wild-type-like) and pale leaves and branches. Seeds were harvested from pods of
phage T7. Note that the transgenes under the control of PT7 have the opposite the pale branches.

Antibacterial activity of chloroplast-produced AMPs diffusion assays (Fig. 7c; see Methods). To determine if removal of
Having established purification procedures for AMPs fused to SUMO, SUMO was required for antimicrobial activity, both the intact fusion
we next wanted to examine, if these proteins have antibacterial activ- proteins and the protease-cleaved proteins were tested. As expected,
ity. To this end, the purified protein samples were tested in radial SUMO-GFP (incubated with or without SUMO protease) did not inhibit

Nature Communications | (2022)13:5856 9

Article https://doi.org/10.1038/s41467-022-33516-1

Fig. 6 | Comparison of transplastomic plants expressing AMPs and fAMPs was performed with anti-HA antibody. The Coomassie-stained membrane is shown
either alone or as fusions to SUMO (S) under the control of different promoter below the immunoblot as a loading control. This blot was performed once, but
systems. a Images of plants grown in sterile culture on synthetic medium. Photos results were confirmed for RAmpER::3-W-5 by an independent blot. Note that
were taken three weeks after transfer of stem cuttings from R::3-W-5 and R::S-3-W-5, SUMO-GFP (SG) accumulates to similarly high levels as the large subunit of
six months after transfer for P::3-W-5 (image from Fig. 2a), and three months after RuBisCO (R) in both Prrn::SUMO-GFP (P::S-GFP) and RAmpER::SUMO-GFP (R::S-
transfer for P::S-3-W-5. Photos of soil-grown plants were taken one month after GFP) plants. Note that the SUMO-GFP protein is not detected in the western blot,
sowing. Scale bars: 5 cm. For images of soil-grown plants after two months, see due to absence of an HA-tag. S25: 25 kDa band of the Spectra™ Multicolor Low
Supplementary Fig. 4. b Northern blot analysis comparing mRNA accumulation Range Protein Ladder (1.7 to 40 kDa; ThermoFisher); P25: 25 kDa band of the
levels in the different transplastomic lines shown in (a) (see Methods section). MB: PageRuler™ Plus Prestained Protein Ladder (10 to 250 kDa; ThermoFisher). RNA
methylene blue staining (as loading control). This blot was performed twice with and protein was extracted from plants grown on synthetic medium to enable
similar results. c Western blot analysis comparing the effects of SUMO fusion on inclusion of lines that did not survive on soil. Source data are provided as a Source
protein accumulation in the different transplastomic lines. Samples of 20 µg total Data file.
leaf protein were separated by gel electrophoresis, blotted and immunodetection

bacteria (Escherichia coli) growth (Fig. 7c). By contrast, purified SUMO- husbandry. In the course of this work, we have explored diverse stra-
HA-WAM1 and SUMO-Novispirin displayed clear antimicrobial activity tegies to utilize transplastomic tobacco plants as production platform
(Fig. 7c). Interestingly, no difference in antimicrobial activity was for antimicrobial peptides (Supplementary Table 2).
observed between the samples treated with and without SUMO pro- Chloroplasts can express foreign proteins to very high levels25,
tease, suggesting that fusion to SUMO does not appreciably affect the including large antibacterial proteins such as phage-derived lytic
activity of AMPs. enzymes (so-called endolysins22,72) and antiviral agents24,73. However,
the expression of very small polypeptides poses serious challenges,
Discussion due to the presence in plastids of specific proteases that degrade
In the face of rapidly spreading resistance of pathogenic bacteria to polypeptides smaller than 65 amino acids36,37. We have addressed this
classical antibiotics, the development of new types of antimicrobial problem by constructing protein fusions that increase AMPs size
agents is becoming one of the most pressing global health challenges above the critical threshold for protease recognition. This was
and, in addition, is of great importance to plant production and animal achieved by fusion of multiple AMPs separated by flexible linkers

Nature Communications | (2022)13:5856 10

Article https://doi.org/10.1038/s41467-022-33516-1

Fig. 7 | Purification and antimicrobial activity of SUMO-AMP fusion proteins

from transplastomic plants. a Ni-NTA agarose purification of SUMO-AMP and
SUMO-GFP fusion proteins based on the 6xHis tag. Protein samples were separated
by gel electrophoresis and blotted. The western blot was stained with Coomassie
prior to blocking and immunodetection with an anti-HA antibody. For extraction,
frozen leaf material was ground and mixed with extraction buffer. An aliquot of this
suspension was directly loaded as total protein (To; incubated with denaturing
sample buffer for 5 min at 95 °C and insoluble material removed by centrifugation).
The insoluble material was then removed by centrifugation, and the supernatant
was used as input for purification (In) by incubation with Ni-NTA agarose beads. The
unbound flow through fraction was collected (Ft), and after washing the column,
the bound SUMO fusion proteins were eluted (E). Asterisks indicate the position of
the fusion proteins in the To and In fractions. Two independent blots were per-
formed with similar results. b Cleavage assay of the purified SUMO fusion proteins
using SUMO protease. 170 µg SUMO-GFP (S-GFP), 45 µg SUMO-HA-WAM1 (S-W),
and 96 µg SUMO-HA-Novispirin (S-N) were incubated with (+) or without (−) 25 units
of SUMO protease (Ulp1: 27 kDa). Of these samples, 35.5 µg SUMO-GFP, 9 µg SUMO-
HA-WAM1, and 20 µg SUMO-HA-Novispirin were separated by SDS-PAGE followed
by western blotting. The resulting membranes were stained with Coomassie fol-
lowed by immunodetection of S-W and S-N with an anti-HA antibody. Two inde-
pendent blots were performed with similar results. c Activity test of purified AMPs
by radial diffusion assays. Equimolar concentrations (150 µM) of SUMO fusions (in
20 mM Tris pH 8.0 buffer) were incubated with or without 2.5 units of SUMO
protease, and applied in an antimicrobial activity radial diffusion assay. Anti-
microbial activity is observed as a clear zone without bacterial growth around the
well. The well in the center of the plate was loaded with the buffer control. Three
independent assays were performed with similar results. For abbreviations, see (b).
Source data are provided as a Source Data file.

expressed in bacteria76. The use of SUMO as carrier protein in chlor-

oplasts is an approach that has not been explored before. Our data
suggest that fusion to SUMO provides a valuable tool for efficient
recombinant protein production and purification from transplastomic
plants. In our study, SUMO fusion had three beneficial effects. First, in
all (f)AMPs tested, we observed improved protein accumulation, pre-
sumably by increasing protein stability. Second, it greatly improved
solubility of the AMPs, thus facilitating purification and bypassing the
need to use detergents for AMP extraction. Third, combined with
lowered expression levels, fusion to SUMO also alleviated the mutant
phenotypes associated with fAMP expression.
Previous work has shown that recombinant proteins that integrate
into or interact with membranes impair chloroplast development, thus
causing pigment deficiency and growth defects35. As AMPs have a
strong propensity to interact with membranes1,3,77, the mutant pheno-
types observed in many of our transplastomic (f)AMP-expressing plants
were not entirely surprising (Figs. 2a, 4a and 6a). We have shown that
the use of an inducible expression system and/or fusion to SUMO can
alleviate these defects, although in most cases, it cannot restore wild-
type-like growth. However, optimization of expression constructs for
several (f)AMPs resulted in plants that display only moderate growth
retardation compared to the wild type (Figs. 4b and 6a). Given that
AMPs are high-value products, such small delays in growth are likely
and/or fusion to SUMO, a soluble protein that can be easily removed acceptable for commercial production. In the non-induced state, the
post-synthesis by cleavage with a highly specific protease. In our work, RAmpER system behaved as a moderately strong promoter resulting in
fusion to the full-length SUMO protein enabled efficient AMP release (f)AMP accumulation with reduced negative impact on plant growth.
by proteolytic cleavage. Although the majority of SUMO fusion was For increased production, induction with theophylline at the optimized
efficiently cleaved, part of the purified product may be in a con- time and concentration could further increase the production17,71.
formation that did not enable recognition by the SUMO protease. An additional product from the research reported here was the
SUMO cleavage mimics SUMO maturation in yeast, where the SUMO generation of a transplastomic line harboring the riboswitch-
protease recognizes the structure of SUMO and cleaves within the controlled T7 RNA polymerase gene (RT7RP line, Fig. 3b). Since this
C-terminal sequence motif GG/ATY converting full-length pre-SUMO line does not contain aadA, the standard marker gene for chloroplast
into the mature form74–76. To obtain a final product without any addi- transformation, it represents an ideal recipient line for super-
tional amino acids, the ATY sequence could also be omitted, in which transformation with transgenes for inducible expression using the
case cleavage takes place between the C-terminal GG of SUMO and the aadA marker and spectinomycin selection (as demonstrated with
fused protein sequence downstream (except if the GG sequence is our series of RAmpER::fAMP transplastomic lines; Figs. 3 and 5; Sup-
followed by a proline residue), as demonstrated for SUMO fusions plementary Figs. 2 and 3; Supplementary Table 2 and Supplementary

Nature Communications | (2022)13:5856 11

Article https://doi.org/10.1038/s41467-022-33516-1

Data 2). This RT7RP recipient line will greatly simplify construct however, for the 15 amino acid linkers, the amino acid sequence had to
assembly in future studies using the RAmpER system for transgene be varied (Supplementary Table 1 and Supplementary Data 2). For
expression in plastids. N-terminal HA-tagging, a single glycine was used as mini linker
In our work, yields of up to 56 µg SUMO-AMP per g leaf material between the initiator methionine and the HA-tag. Glycine was selected
were obtained, corresponding to 18 µg of HA-WAM1 and 12 µg of HA- as penultimate amino acid based on previous success with the
Novispirin. Previous studies on the production of cationic AMPs from expression of griffithsin24. Nucleotide sequences of 15 nt identical to
the plastid genome reported production of AMP fusions to GFP with the flanking regions of the insertion site in the chloroplast transfor-
yields of 5 µg per g fresh leafy biomass (chromatography purification) mation vectors were included in the synthesized DNA to facilitate
or 53 µg per g (organic extraction) for a Retrocyclin-GFP fusion and homology-based cloning.
8 µg per g for a Protegrin-1-GFP fusion26. Production of cationic AMPs
in plants by stable genetic transformation of the nuclear genome Generation of plastid transformation constructs
yielded up to 7 µg per g fresh weight (for MsrA227,28). Production of To achieve high-level constitutive expression of AMP constructs,
AMPs on the surface of a recombinant plant virus that was used to plasmid pDK323 (Fig. 1a) was used to place the genes of interest under
infect plants yielded up to 25 µg/g fresh weight of AMP (for SP1-127,29). the control of the tobacco plastid ribosomal RNA operon promoter
By contrast, anionic AMPs fused to the elastin protein and expressed (Prrn) fused to the T7 phage gene10 leader (g10L) and the 3′ UTR of the
transiently in plants appear to accumulate to higher levels than chloroplast atpA gene from Chlamydomonas reinhardtii (CrTatpA).
cationic AMPs, reaching yields of up to 563 µg/g FW, corresponding to pDK323 also contains the spectinomycin resistance gene aadA
113 µg/g FW of AMP (for ADP227,32). However, cationic AMPs are more that encodes the enzyme aminoglycoside 3′′-adenylyltransferase65 and
versatile antimicrobial agents than anionic AMPs, because they have a is driven by the C. reinhardtii psaA promoter. In addition, the vector
stronger binding affinity to microbial membranes, due to the abundant contains flanking homology regions for integration into the plastid
presence of anionic compounds in the membranes of bacteria and genome between the trnfM and trnG genes64 by homologous recom-
fungi (including lipopolysaccharides in gram-negative bacteria, lipo- bination (Fig. 1).
teichoic acid in gram-positive bacteria and mannans in fungi78). For inducible expression from the plastid genome, plasmid
In our work, we demonstrated activity of plastid-produced HA- pMPH36 was created. pMPH36 is similar to pDK323 that was used for
WAM1 and HA-Novispirin against the gram-negative bacterium E. coli. constitutive expression, but contains the T7 promoter instead of the
Previous studies had demonstrated that WAM1 and Novispirin are Prrn promoter and harbors the T7 promoter-driven transgene in
active against both gram-negative and gram-positive bacteria56,79 opposite orientation to trnG (see Figs. 1, 3 and 5 for construct maps).
(Supplementary Data 1), and given the low folding requirements of pMPH36 also contains flanking regions for integration between trnfM
AMPs, their broad activity is expected to be the preserved in the and trnG in the plastid genome by homologous recombination. Con-
plastid-produced peptides. A particularly interesting finding from our structs derived from pMPH36 were used to supertransform pDK320
antibacterial activity assays was that the bactericidal activity of AMPs (Fig. 3b) transplastomic recipient plants that carried a riboswitch-
was independent of SUMO removal. This observation suggests that controlled T7 RNA polymerase gene17 (RT7RP) in the plastid genome
folding of AMPs into their active three-dimensional conformation is between the genes ycf3 and psaA. RT7RP transplastomic plants were
not affected by fusion to a comparatively large N-terminal protein generated by using the aphA6 gene84 (encoding aminoglycoside
domain. It also offers the possibility to use AMPs as fusions to other phosphotransferase) providing kanamycin resistance as selection
functional proteins in practical applications, thus avoiding the need for marker (Fig. 3). This strategy allowed the use of aadA as marker for
protease cleavage and providing the option to work with proteins that subsequent supertransformation of RT7RP plants with pMPH36-
are more manageable in size, multifunctional and potentially more derived constructs, resulting in plants with a functional RNA
stable. amplification-enhanced riboswitch (RAmpER) system17. Some expres-
AMPs are also naturally produced by plants and have been pro- sion cassettes controlled by the T7 promoter were additionally used
posed to constitute an important part of plant immunity80–82. Future for transformation of wild-type plants.
work will be directed towards the construction of tighter inducible DNA sequences for the different AMPs were synthesized (Gene-
expression systems that allow the generation of AMP-synthesizing Cust) and used as template for PCR. PCR products were generated with
plants with no negative impact on plant growth. In addition to pro- Phusion High-Fidelity DNA Polymerase (Thermo Fisher Scientific), gel
viding a potential source of AMPs for pharmaceutical purposes, such purified (Nucleospin® Gel and PCR Clean-up kit; Macherey-Nagel) and
plants could be tested for resistance to plant-specific bacterial and inserted into EcoRV linearized vectors (pDK323, pMPH36) using In-
fungal pathogens, potentially providing a powerful tool for crop pro- Fusion® HD Cloning (Takara) (Figs. 1, 3 and 5). To form larger fusions of
tection against diseases in agriculture. six or nine AMPs, different constructs comprising three peptides each
were joined by overlap-extension PCR followed by In-Fusion® HD
Methods Cloning.
Design of expression constructs for fusions of antimicrobial To construct SUMO fusions, the DNA sequence encoding the full-
peptides length (except for the last amino acid) yeast SUMO (SMT3; NCBI
The selected AMP sequences were linked to each other with either 5 or Reference Sequence: NP_010798.1) with an N-terminal 6xHis tag
15 amino acid linkers composed of glycine and serine. An N-terminal (MEHHHHHHGG-) was codon optimized according to the preferred
human influenza hemagglutinin (HA) tag (MGYPYDVPDYA) and a codon usage in chloroplasts63 and synthesized (Invitrogen). The
C-terminal 6xHis-tag were added, also connected by 5 or 15 amino acid sequence encoding SUMO ended with a SnaBI restriction site which
linkers. DNA sequences encoding these fused AMP amino acid was used for plasmid linearization. A glutamate residue was included
sequences were generated by digital reverse translation and optimized between the initiator methionine and the 6xHis-tag (Supplementary
to the preferred codon usage of tobacco chloroplasts63. The DNA Data 2), because glutamate as penultimate amino acid was previously
sequences encoding the linkers were adjusted to the codon usage in shown to increase protein stability in chloroplasts85. The DNA fragment
tobacco chloroplasts63, in such a way that any repetition of 10 or more was inserted into the chloroplast transformation vectors (pDK323 and
nucleotides between the different linkers was avoided (Supplementary pMPH36) by EcoRV linearization and In-Fusion® HD homology-based
Table 1 and Supplementary Data 2), to reduce the risk of unwanted integration, resulting in SUMO expression vectors pMPH50 and
recombination within the chloroplast genomes61,83. For the 5 amino pMPH71. PCR products encoding the proteins to be SUMO tagged and
acid linkers, this was possible with the amino acid sequence GGSGG, a short sequence restoring the codons for the last two amino acids of

Nature Communications | (2022)13:5856 12

Article https://doi.org/10.1038/s41467-022-33516-1

SUMO were inserted into the linearized transformation vectors by In- RNA gel blot analyses
Fusion® HD homology-based integration. Transgene expression in transplastomic lines was analyzed by north-
To obtain SUMO-GFP fusions, the synthetic green fluorescent ern blotting. Total plant RNA was extracted using TRIzol® (Thermo
protein (sGFP) was used86. The sGFP DNA sequence used is identical to Fisher Scientific) following the protocol of the manufacturer, and
NCBI GenBank entry DQ370424.1 (NCBI protein ID GenBank: quantified by optical density measurements using a Thermo Scien-
ABD16415.1), but lacks the codons encoding the last two amino acids (I tific™ NanoDrop™ spectrophotometer. Samples of 3 µg total RNA were
and L), and ends with the stop codon TAA. denatured at 75 °C for 15 min and then separated by agarose gel elec-
trophoresis. Denaturing 1% (w/v) agarose gels were prepared in 1x
Chloroplast transformation MOPS buffer [0.1 M 3-(N-morpholino) propanesulfonic acid, 0.3 M
Vectors for chloroplast transformation were amplified in E. coli strain NaOAc, 1 mM EDTA; pH 7] in presence of 16% (v/v) formaldehyde
TOP 10, and plasmid DNA was purified with anion exchange columns (Sigma). Separated RNA was transferred onto Hybond-N nylon mem-
(Macherey-Nagel). The plasmid preparations were resequenced prior branes (GE Healthcare) by overnight capillary blotting using 5 x SSC
to plant transformation to confirm absence of mutations from the buffer (0.75 M NaCl, 75 mM trisodium citrate; pH 7), and covalently
transgenes or the flanking chloroplast homology regions. bound to the membrane by exposure to UV light (0.12 J/cm2; UV-
Samples of 30 µg vector DNA were coated onto gold particles and crosslinker BLX-254, Vilber Lourmat). Blotted nylon membranes were
shot by particle bombardment into young Nicotiana tabacum cv. Petit stained with 0.03% methylene blue (SERVA) solution in 0.3 M NaOAc,
Havana leaves harvested from plants cultivated under sterile condi- pH 5.2, and scanned using an EPSON Perfection V700 Photo scanner. A
tions. Bombarded leaves were cut into pieces and incubated on SpeI/EcoRV restriction fragment digested from vector pDK323 (cor-
regeneration medium for tobacco shoot organogenesis composed of responding to the Chlamydomonas reinhardtii atpA 3′ UTR) was
MS elements + modified vitamins (Duchefa M0245), sucrose (30 g/L), labelled with [α-32P]-dCTPs (Hartmann Analytic GmbH) using the
sorbitol (18.22 g/L), mannitol (18.22 g/L), 6-benzylaminopurine (BAP; Amersham™ Megaprime™ DNA Labeling System (Cytiva) following the
1 mg/L), 1-naphthaleneacetic acid (NAA; 0.1 mg/L), agar (5.4 g/L) at pH manufacturer’s protocol. Hybridization was performed overnight at
5.8, and supplemented with 500 mg/mL spectinomycin (Duchefa). 65 °C in Church buffer [1% (w/v) BSA, 0.5 M Na2HPO4 (pH 7.2), 7% (w/v)
Primary spectinomycin-resistant shoots were subjected to additional SDS, 1 mM EDTA (pH 8.0)]. Membranes were exposed to a storage
regeneration rounds to select for homoplasmy. To this end, callus or phosphor screen (GE Healthcare) followed by radioactive signal
leaf pieces of antibiotic-resistant lines were transferred to fresh detection using the Typhoon™ TRIO + scanner (GE Healthcare). For
regeneration medium with 500 mg/mL spectinomycin. In parallel, tis- images of uncropped blots, see Source Data file.
sue samples were incubated on regeneration medium containing
500 mg/mL streptomycin (Duchefa), to distinguish true transplas- Plant growth under sterile conditions
tomic clones from spontaneous spectinomycin-resistant mutants67. Plants were grown under aseptic conditions in plastic boxes (Magenta)
Finally, emerging homoplasmic shoots were rooted on Murashige & on MS medium with 3% sucrose (with or without 500 mg/L spectino-
Skoog (MS) medium87 with sucrose (3%) and 500 mg/mL spectino- mycin) under a diurnal cycle of 50 µmol m−2 s−1 light for 16 h at 25 °C,
mycin, transferred to soil, and grown to maturity. Per construct, at and 8 h darkness at 20 °C. Plants that did not produce seeds in the
least three independent transplastomic lines were created. greenhouse were maintained in sterile culture by rooting of stem
cuttings every three to six months. Pigment-deficient plants were
DNA gel blot analysis maintained at minimal light intensities of 3–5 µmol m−2 s−1.
The complete replacement of the wild-type plastid genome with the
transgenic genome (homoplasmy) was confirmed by restriction frag- Inheritance assays
ment length polymorphism analysis (RFLP) via Southern blotting. To confirm the homoplasmic state of transplastomic plants, surface
Briefly, DNA was isolated from snap-frozen leaf material by a cetyl- sterilized seeds were germinated on antibiotic-containing MS medium
trimethylammoniumbromide (CTAB)-based DNA extraction method88. with 3% sucrose. Antibiotic-resistant seedlings were identified by
DNA concentrations were determined with a Thermo Scientific™ selection in the presence of spectinomycin (500 µg/mL).
NanoDrop™ spectrophotometer. Samples of 3 µg DNA were digested
in a 50 µL volume overnight at 37 °C with the restriction enzyme BglII Greenhouse growth of plants
(NEB). The digest was subsequently mixed with 10 µL 6x DNA gel Tobacco plants (Nicotiana tabacum var. Petit Havana) were grown in
loading dye (Thermo Fisher Scientific). Samples of 30 µL were elec- soil in the greenhouse with light supplementation. Artificial light was
trophoretically separated in a 0.8% agarose gel. The gel was incubated used from 6 am to 10 pm (i.e., for 16 h). The measured total average light
for 15 min in depurination buffer (0.25 M HCl), rinsed with ddH2O and intensity was (PPFD)/(PAR) 195 µmol m-2 s-1, the day temperature set to
incubated with 0.5 M NaOH for 30 min, rinsed with ddH2O and incu- 25 °C, the night temperature 20 °C, and the relative humidity set to 55%.
bated in 0.5 M NaOH and 1.5 M NaCl for 30 min, and finally rinsed with
ddH2O and incubated in 1 M Tris, 3 M NaCl for 15 min. DNA gels were Grafting
blotted with 10 x SSC buffer (1.5 M NaCl, 0.15 M trisodium citrate) onto To allow growth on soil and seed production of AMP-expressing plants
a Hybond-XL membrane (GE Healthcare), followed by UV crosslinking that displayed severe pigment-deficient phenotypes and did not grow
(0.12 J/cm2; UV-crosslinker BLX-254, Vilber Lourmat). Restriction frag- autotrophically, transplastomic shoots were grafted onto wild-type
ments were detected by hybridization with a [α-32P]-dCTP-labelled shoots in tissue culture. Autoclaved pieces of silicon tubing, cut open
psaB probe generated by PCR (primers: CCCAGAAAGAGGCTGGCCC longitudinally to allow later removal, were used to join wild-type stock
and CCCAAGGGGCGGGAACTGC). Overnight hybridization was per- and transplastomic scion. Successful grafts were transferred to soil
formed in Church buffer [1% (w/v) BSA, 0.5 M Na2HPO4, pH 7.2, 7% (w/ and grown in the greenhouse. Wild-type side shoots were pruned every
v) SDS, 1 mM EDTA, pH 8] at 65 °C. Membranes were briefly rinsed and 14 to 30 days to prevent overgrowth and encourage growth and
incubated with 2 x SSC and 0.1% SDS at 25 °C for 20 min and incubated flowering of the mutant shoot.
twice with 0.5 x SSC and 0.1% SDS at 65 °C for 15 min. Membranes were
exposed to a storage phosphor screen (GE Healthcare) followed by Total protein extraction
imaging of the screen with a Typhoon™ TRIO + scanner (GE Health- To extract total leaf protein from tobacco plants, leaf pieces were snap
care). For images of blots, see Supplementary Figs. 1–3, and for frozen in liquid nitrogen in 2 mL Eppendorf tubes and pulverized with
uncropped blots, see Source Data file. 5 mm steel balls for one minute and 15 s at 20 Hz in a Retsch mill. Total

Nature Communications | (2022)13:5856 13

Article https://doi.org/10.1038/s41467-022-33516-1

protein was isolated by a phenol-based extraction protocol89. Protein 6.8, 8% SDS, 0.2 M DTT, 0.02% Coomassie Brilliant Blue G-250),
pellets were dissolved in 1% sodium dodecyl sulfate (SDS). Protein denatured for 5 min at 95 °C, centrifuged for 1 min at 15,000 x g and
concentrations were quantified with the Pierce BCA Protein Assay Kit then loaded onto the gel. Gels were run overnight at 16 °C at constant
(Thermo Fisher Scientific) and measured with a CLARIOstar plate voltage.
reader (BMG Labtech) using a bovine serum albumin (BSA) dilution For western blotting, protein gels were electroblotted to Hybond-
series as standard. Calculations of protein concentrations based on P PVDF membranes (GE Healthcare) in transfer buffer (192 mM glycine,
spectrophotometric data were done with Microsoft Excel (Microsoft 25 mM Tris, pH 8.3) in a Trans-Blot cell (Bio-Rad) for 2 h at 1.0 A at 16 °C.
365 version 2202). Blotted membranes were stained with Coomassie prior to
immunochemical detection. To this end, the membranes were incu-
His-tag-based protein purification bated with Coomassie 250 R (50 mg per 200 mL in 50% methanol and
Leaf material was snap frozen in liquid nitrogen, ground with a Retsch 7% acetic acid) directly after transfer. Membranes were stained for
mill in metal containers for one minute at 25 Hz, and incubated with 2–5 min until the membrane was dark blue, and subsequently de-
extraction buffer24 (100 mM Tris, 300 mM NaCl, 20 mM ascorbic acid, stained with 50% methanol and 7% acetic acid for one hour or until the
10 mM sodium metabisulfite, 10 mM imidazole, final pH 8.0). 100 µL background was sufficiently reduced. The membranes were then
buffer was used per 100 mg leaf material, and the samples were incu- rinsed briefly with ddH2O and scanned. The Coomassie was subse-
bated on ice for approximately 30 min with regular mixing by vor- quently removed with 100% methanol by incubating for up to one
texing. Insoluble material was pelleted by centrifugation (16,000 x g hour, and the membranes were washed with Tris-Buffered Saline (TBS)
for 30 min at 4 °C). The supernatant was used as input for purification. with 1% TWEEN® 20 (v/v; TBS-T) before proceeding with the blocking
For large-scale purification, frozen leaf material was ground with buffer procedure prior to immuno-detection.
(3 mL per g leaf material) in a countertop blender (two short pulses of Blotted membranes were blocked with 5% milk powder (w/v) in
10 s at low speed, followed by three rounds of grinding for 60 s at high TBS-T for one hour at room temperature. The membranes were then
speed). Tissue debris and other insoluble material were pelleted in rinsed twice with TBS-T and subsequently washed once for 15 min and
large beakers by centrifugation (15,000 x g for 30 min at 4 °C). The twice for 5 min in TBS-T on a shaker. The blots were subsequently
supernatant (lysate) was decanted through filter papers, and the incubated with the primary antibody (anti-HA; GenScript THETM HA-tag
resulting filtrate used as input for purification. Antibody, mAb, mouse, catalogue No: A01244-100; 0.5 µg/mL) for one
Nickel-nitrilotriacetic acid (Ni-NTA) agarose beads (QIAGEN) were hour at room temperature or overnight at 4 °C. Primary antibodies
resuspended to a slurry in the storage solution. The slurry volume were prepared in TBS-T with 200 µL Micro-O-Protect (80% ethanol,
required for purification was transferred into an Eppendorf tube, the 7.5% 5-bromo-5-nitro-1,3-dioxane, 7.5% 2-methyl-4-isothiazolin-3-one
beads were pelleted by centrifugation and the supernatant discarded. hydrochloride) and stored at 4 °C. After incubation with the antibody
The beads were then equilibrated by washing once with 2.5 volumes of solution, the membranes were washed as described above, and incu-
ddH2O and twice with 2.5 volumes of extraction buffer. Pelleting by bated with HRP-conjugated anti-mouse antibody (Sigma A9044; dilu-
centrifugation was performed at 700 x g for 10–15 min at 4 °C. For small ted freshly in TBS-T 1:10.000). Afterwards, the membranes were
batch purification, bead slurry equivalent to 10% or 60% of the volume washed again, and the signals were detected with the enhanced che-
of the protein extract was used for subsequent protein purifications. miluminescence kit (ECL® PLUS system, GE Healthcare). Images of
For purification, beads were incubated with protein extract by uncropped blots are available in the Source Data file.
slow rotation for one hour, washed twice with 2.5 slurry volumes of
washing buffer (extraction buffer plus 20 mM imidazole), and eluted Antimicrobial activity assays in vitro
with 25 or 33% of the protein extract input volume of elution buffer A modified radial diffusion assay92–94 was used to test antimicrobial
(extraction buffer with 250 mM imidazole). All purification steps were activity of protein extracts. E. coli TOP 10 cells were inoculated over-
performed at 4 °C. night at 37 °C in liquid casein peptone soybean flour peptone (CASO)
For large-scale batch purifications, equilibrated Ni-NTA agarose medium. 10 µL samples of the overnight cultures were inoculated in
beads (2% of the lysate volume) were incubated in the supernatant with 10 mL CASO, and grown for 3 h. The cultures were then pelleted by
a magnetic stirrer bar for one hour on ice. Subsequently, the lysate centrifugation (3000 x g, 10 min, 20 °C) and the pellets were rinsed
with the beads was decanted in several batches onto gravity columns carefully with cold (4 °C) 10 mM sodium phosphate buffer, pH 7.4
(QIAGEN 5 mL polypropylene columns), allowing the Ni-NTA agarose (NAPB), followed by resuspension in 5 mL NAPB. The OD620 was
beads to sediment on the filter while letting the lysate to flow through. measured in the suspension and a 10x diluted suspension, and the
The columns were then washed 3 times with 10 mL of washing buffer number of colony-forming units (CFU) was determined by the fol-
(for higher purity) or extraction buffer (for higher yield), and eluted 4 lowing equation 1.
times with 1 mL of elution buffer.
Eluted fractions (3.5 mL of 4 mL) were loaded on a PD-10 Desalting CFU=mL = OD620 × 2:5 × 108 ð1Þ
Column (GE Healthcare), starting with the most concentrated frac-
tions, buffer-exchanged into 20 mM Tris/HCl pH 8.0, and con- The volume of bacterial suspension containing 4 × 106 CFU was
centrated by ultrafiltration (Amicon® Ultra‐4 centrifugal filters; Merck) calculated and added to 10 mL of melted (55 °C) bottom layer medium.
with a 3000 Nominal Molecular Weight Limit (NMWL) cut‐off. Protein The liquid bottom layer medium was mixed with the bacteria by
concentrations were determined with the Pierce BCA Protein Assay Kit quickly inverting the tube 10 times, and poured into a Petri dish (8.5 cm
(Thermo Fisher Scientific) and measured with a CLARIOstar plate diameter). The bottom layer medium was composed of 10 mM sodium
reader (BMG Labtech) using a dilution series of BSA as standard. Pro- phosphate buffer (NAPB), pH 7.4, 100x diluted CASO liquid medium,
tein concentrations based on spectrophotometric data were calcu- and 1% agarose (for nucleic acid electrophoresis, Sigma).
lated with Microsoft Excel (Microsoft 365 version 2202). Protein samples to be assayed were loaded into wells in the bot-
tom layer. Wells were made with 3 mL syringes (Braun) by pressing the
Tris-tricine SDS-PAGE and immunoblotting tip (without an attached needle) into the agar, bringing up the plunger
To analyze protein accumulation and follow protein purification, slightly and pulling the syringe up quickly. The resulting wells were
protein extracts were separated by electrophoresis in 10% Tris-tricine approximately 4 mm in diameter. Samples (of 10 µL) were loaded into
SDS-polyacrylamide gels90,91. Protein extracts were mixed with equal the wells in the agarose, and allowed to diffuse for two to three hours
volumes of 2 x Tris-tricine sample buffer (24% glycerin, 0.1 M Tris pH at 37 °C. Subsequently, 10 mL nutritious top layer (2x CASO medium

Nature Communications | (2022)13:5856 14

Article https://doi.org/10.1038/s41467-022-33516-1

with 1% agarose) was poured on top of the bottom layer. After solidi- 15. Cai, Y. et al. Griffithsin with A broad-spectrum antiviral activity by
fication of the top layer, plates were incubated upside down overnight binding glycans in viral glycoprotein exhibits strong synergistic
at 37 °C, and inspected the next morning for zones without bacterial effect in combination with a pan-coronavirus fusion inhibitor tar-
growth surrounding the wells. geting SARS-CoV-2 Spike S2 subunit. Virologica Sin. 35,
857–860 (2020).
SUMO digestion 16. Vamvaka, E. et al. Rice endosperm is cost-effective for the pro-
SUMO protease digestion was performed according to a published duction of recombinant griffithsin with potent activity against HIV.
protocol76 with some modifications. SUMO protease (2500 units Plant Biotechnol. J. 14, 1427–1437 (2016).
6xHis-tagged Ulp1; Sigma Aldrich, SAE0067-2500UN) was diluted to 25 17. Emadpour, M., Karcher, D. & Bock, R. Boosting riboswitch efficiency
units per µL in 50% glycerol with 1 mM DTT, aliquots were snap frozen by RNA amplification. Nucleic Acids Res. 43, e66 (2015).
in liquid nitrogen and stored at −20 °C until use. Digestion was per- 18. Lössl, A. et al. Inducible trans-activation of plastid transgenes:
formed in digestion buffer (20 mM Tris/HCl, pH 8.0) for 2–3 h at room Expression of the R. eutropha phb operon in transplastomic
temperature (25 °C). SUMO digestion of fusion proteins was per- tobacco. Plant Cell Physiol. 46, 1462–1471 (2005).
formed with a final concentration of 0.25 to 0.83 units per µL, and a 19. Rojas, M., Yu, Q., Williams-Carrier, R., Maliga, P. & Barkan, A. Engi-
total amount of 2.5 to 25 units per 25 to 170 µg of protein substrate. neered PPR proteins as inducible switches to activate the expres-
Cleavage efficiency was determined by image quantification of western sion of chloroplast transgenes. Nat. Plants 5, 505–511 (2019).
blots with ImageJ version v1.52r (https://imagej.nih.gov/ij/). 20. Egelkrout, E., Rajan, V. & Howard, J. A. Overproduction of recom-
binant proteins in plants. Plant Sci. 184, 83–101 (2012).
Reporting summary 21. Werner, S., Breus, O., Symonenko, Y., Marillonnet, S. & Gleba, Y.
Further information on research design is available in the Nature High-level recombinant protein expression in transgenic plants by
Research Reporting Summary linked to this article. using a double-inducible viral vector. Proc. Natl Acad. Sci. 108,
14061–14066 (2011).
Data availability 22. Oey, M., Lohse, M., Kreikemeyer, B. & Bock, R. Exhaustion of the
A reporting summary for this Article is available as a Supplementary chloroplast protein synthesis capacity by massive expression of a
Information file. Data supporting the findings of this work are available highly stable protein antibiotic. Plant J. 57, 436–445 (2009).
within the paper and its Supplementary Information files. Source data 23. Bock, R. & Warzecha, H. Solar-powered factories for new vaccines
are provided with this paper. and antibiotics. Trends Biotechnol. 28, 246–252 (2010).
24. Hoelscher, M. et al. High-level expression of the HIV entry
References inhibitor griffithsin from the plastid genome and retention of bio-
1. Nguyen, L. T., Haney, E. F. & Vogel, H. J. The expanding scope of logical activity in dried tobacco leaves. Plant Mol. Biol. 97,
antimicrobial peptide structures and their modes of action. Trends 357–370 (2018).
Biotechnol. 29, 464–472 (2011). 25. Bock, R. Engineering plastid genomes: methods, tools, and appli-
2. Zasloff, M. Antimicrobial peptides of multicellular organisms. Nat- cations in basic research and biotechnology. Annu. Rev. Plant Biol.
ure 415, 389–395 (2002). 66, 211–241 (2015).
3. Lazzaro, B. P., Zasloff, M. & Rolff, J. Antimicrobial peptides: Appli- 26. Lee, S. B., Li, B., Jin, S. & Daniell, H. Expression and characterization
cation informed by evolution. Science 368, eaau5480 (2020). of antimicrobial peptides Retrocyclin-101 and Protegrin-1 in chlor-
4. Wang, G., Li, X. & Wang, Z. APD3: The antimicrobial peptide data- oplasts to control viral and bacterial infections. Plant Biotechnol. J.
base as a tool for research and education. Nucleic Acids Res. 44, 9, 100–115 (2011).
D1087–D1093 (2016). 27. Shanmugaraj, B., Bulaon, C. J. I., Malla, A. & Phoolcharoen, W. Bio-
5. Fjell, C. D., Hiss, J. A., Hancock, R. E. W. & Schneider, G. Designing technological insights on the expression and production of anti-
antimicrobial peptides: form follows function. Nat. Rev. Drug Dis- microbial peptides in plants. Molecules 26, (2021).
cov. 11, 37–51 (2012). 28. Yevtushenko, D. P. & Misra, S. Comparison of pathogen-induced
6. Smith, P. A. & Romesberg, F. E. Combating bacteria and drug expression and efficacy of two amphibian antimicrobial peptides,
resistance by inhibiting mechanisms of persistence and adaptation. MsrA2 and temporin A, for engineering wide-spectrum disease
Nat. Chem. Biol. 3, 549–556 (2007). resistance in tobacco. Plant Biotechnol. J. 5, 720–734 (2007).
7. Rodrigues, G., Maximiano, M. R. & Franco, O. L. Antimicrobial 29. Zeitler, B. et al. Production of a de-novo designed antimicrobial
peptides used as growth promoters in livestock production. Appl. peptide in Nicotiana benthamiana. Plant Mol. Biol. 81,
Microbiol. Biotechnol. 105, 7115–7121 (2021). 259–272 (2013).
8. Gidalevitz, D. et al. Interaction of antimicrobial peptide protegrin 30. Paškevičius, Š. et al. Plant-expressed pyocins for control of Pseu-
with biomembranes. Proc. Natl Acad. Sci. 100, 6302–6307 (2003). domonas aeruginosa. PLoS One 12, e0185782 (2017).
9. Ding, L. et al. Interaction of antimicrobial peptides with lipopoly- 31. Łojewska, E. et al. Production of recombinant colicin M in Nicotiana
saccharides. Biochemistry 42, 12251–12259 (2003). tabacum plants and its antimicrobial activity. Plant Biotechnol. Rep.
10. Parachin, N. S. et al. Expression systems for heterologous produc- 14, 33–43 (2020).
tion of antimicrobial peptides. Peptides 38, 446–456 (2012). 32. Ghidey, M., Islam, S. M. A., Pruett, G. & Kearney, C. M. Making plants
11. Li, Y. & Chen, Z. RAPD: A database of recombinantly-produced into cost-effective bioreactors for highly active antimicrobial pep-
antimicrobial peptides. FEMS Microbiol. Lett. 289, 126–129 (2008). tides. N. Biotechnol. 56, 63–70 (2020).
12. Ma, J. K.-C. et al. Molecular farming for new drugs and vaccines. 33. Li, Y. Recombinant production of antimicrobial peptides in
EMBO Rep. 6, 593–599 (2005). Escherichia coli: A review. Protein Expr. Purif. 80, 260–267 (2011).
13. Lobato Gómez, M. et al. Contributions of the international plant 34. Giomarelli, B. et al. Recombinant production of anti-HIV protein,
science community to the fight against human infectious diseases – griffithsin, by auto-induction in a fermentor culture. Protein Expr.
part 1: epidemic and pandemic diseases. Plant Biotechnol. J. 19, Purif. 47, 194–202 (2006).
1901–1920 (2021). 35. Scotti, N. et al. The HIV-1 Pr55gag polyprotein binds to plastidial
14. Tusé, D., Tu, T. & McDonald, K. A. Manufacturing economics of membranes and leads to severe impairment of chloroplast bio-
plant-made biologics: case studies in therapeutic and industrial genesis and seedling lethality in transplastomic tobacco plants.
enzymes. Biomed. Res. Int. 2014, 256135 (2014). Transgenic Res. 24, 319–331 (2015).

Nature Communications | (2022)13:5856 15

Article https://doi.org/10.1038/s41467-022-33516-1

36. Kmiec, B. et al. Organellar oligopeptidase (OOP) provides a com- with phospholipid membranes. Biochemistry 31,
plementary pathway for targeting peptide degradation in mito- 12416–12423 (1992).
chondria and chloroplasts. Proc. Natl Acad. Sci. 110, 55. Navon-Venezia, S., Feder, R., Gaidukov, L., Carmeli, Y. & Mor, A.
E3761–E3769 (2013). Antibacterial properties of Dermaseptin S4 derivatives with in vivo
37. Moberg, P. et al. Characterization of a novel zinc metalloprotease activity. Antimicrob. Agents Chemother. 46, 689–694 (2002).
involved in degrading targeting peptides in mitochondria and 56. Sawai, M. V. et al. Impact of single-residue mutations on the
chloroplasts. Plant J. 36, 616–628 (2003). structure and function of ovispirin/novispirin antimicrobial pep-
38. Wollman, F. An antimicrobial origin of transit peptides accounts for tides. Protein Eng. 15, 225–232 (2002).
early endosymbiotic events. Traffic 17, 1322–1328 (2016). 57. Wang, J. et al. Ancient antimicrobial peptides kill Antibiotic-
39. Hager, M., Biehler, K., Illerhaus, J., Ruf, S. & Bock, R. Targeted Resistant pathogens: Australian mammals provide new options.
inactivation of the smallest plastid genome-encoded open reading PLoS One 6, 1–8 (2011).
frame reveals a novel and essential subunit of the cytochrome b6f 58. Yang, X., Lee, W. H. & Zhang, Y. Extremely abundant antimicrobial
complex. EMBO J. 18, 5834–5842 (1999). peptides existed in the skins of nine kinds of Chinese odorous frogs.
40. Yan, H. & Hancock, R. E. W. Synergistic interactions between J. Proteome Res. 11, 306–319 (2012).
mammalian antimicrobial defense peptides. Antimicrob. Agents 59. Chen, X., Zaro, J. L. & Shen, W. C. Fusion protein linkers: Property,
Chemother. 45, 1558–1560 (2001). design and functionality. Adv. Drug Deliv. Rev. 65,
41. Dong, B., Cheng, R. Q., Liu, Q. Y., Wang, J. & Fan, Z. C. Multimer of 1357–1369 (2013).
the antimicrobial peptide Mytichitin-A expressed in Chlamydomo- 60. Iamtham, S. & Day, A. Removal of antibiotic resistance genes from
nas reinhardtii exerts a broader antibacterial spectrum and transgenic tobacco plastids. Nat. Biotechnol. 18, 1172–1176 (2000).
increased potency. J. Biosci. Bioeng. 125, 175–179 (2018). 61. Rogalski, M., Ruf, S. & Bock, R. Tobacco plastid ribosomal protein
42. Zhang, M. et al. Expression of a recombinant hybrid antimicrobial S18 is essential for cell survival. Nucleic Acids Res. 34,
peptide magainin II-cecropin B in the mycelium of the medicinal 4537–4545 (2006).
fungus Cordyceps militaris and its validation in mice. Microb. Cell 62. Kuroda, H. & Maliga, P. Complementarity of the 16S rRNA penulti-
Fact. 17, 1–14 (2018). mate stem with sequences downstream of the AUG destabilizes the
43. Panavas, T., Sanders, C. & Butt, T. R. SUMO Fusion Technology for plastid mRNAs. Nucleic Acids Res. 29, 970–975 (2001).
Enhanced Protein Production in Prokaryotic and Eukaryotic 63. Shimada, H. & Sugiura, M. Fine structural features of the chloroplast
Expression Systems. in SUMO Protocols (ed. Ulrich, H. D.) 303–317 genome: comparison of the sequenced chloroplast genomes.
(Humana Press). https://doi.org/10.1007/978-1-59745-566-4_ Nucleic Acids Res. 19, 983–995 (1991).
20 (2009). 64. Ruf, S., Hermann, M., Berger, I. J., Carrer, H. & Bock, R. Stable
44. Bommarius, B. et al. Cost-effective expression and purification of genetic transformation of tomato plastids and expression of a for-
antimicrobial and host defense. Pept. Escherichia coli. Pept. 31, eign protein in fruit. Nat. Biotechnol. 19, 870–875 (2001).
1957–1965 (2011). 65. Svab, Z. & Maliga, P. High-frequency plastid transformation in
45. Gao, X. et al. Soluble cytoplasmic expression, rapid purification, tobacco by selection for a chimeric aadA gene. Proc. Natl Acad. Sci.
and characterization of cyanovirin-N as a His-SUMO fusion. Appl. 90, 913–917 (1993).
Microbiol. Biotechnol. 85, 1051–1060 (2010). 66. Greiner, S. et al. Chloroplast nucleoids are highly dynamic in ploidy,
46. Feng, X. et al. Recombinant expression, purification, and anti- number, and structure during angiosperm leaf development. Plant
microbial activity of a novel hybrid antimicrobial peptide LFT33. J. 102, 730–746 (2020).
Appl. Microbiol. Biotechnol. 95, 1191–1198 (2012). 67. Bock, R. Transgenic plastids in basic research and plant bio-
47. He, J. et al. Novel synthetic antimicrobial peptides against Strep- technology. J. Mol. Biol. 312, 425–438 (2001).
tococcus mutans. Antimicrob. Agents Chemother. 51, 68. Zhou, F., Karcher, D. & Bock, R. Identification of a plastid inter-
1351–1358 (2007). cistronic expression element (IEE) facilitating the expression of
48. Hiemstra, P. S., van den Barselaar, M. T., Roest, M., Nibbering, P. H. & stable translatable monocistronic mRNAs from operons. Plant J. 52,
van Furth, R. Ubiquicidin, a novel murine microbicidal protein pre- 961–972 (2007).
sent in the cytosolic fraction of macrophages. J. Leukoc. Biol. 66, 69. Lu, Y., Rijzaani, H., Karcher, D., Ruf, S. & Bock, R. Efficient metabolic
423–428 (1999). pathway engineering in transgenic tobacco and tomato plastids
49. Cole, A. M. et al. Cutting Edge: IFN-Inducible ELR− CXC chemokines with synthetic multigene operons. Proc. Natl Acad. Sci. 110,
display defensin-like antimicrobial activity. J. Immunol. 167, E623–E632 (2013).
623–627 (2001). 70. Verhounig, A., Karcher, D. & Bock, R. Inducible gene expression
50. Yang, D. et al. Many chemokines including CCL20/MIP-3alpha dis- from the plastid genome by a synthetic riboswitch. Proc. Natl Acad.
play antimicrobial activity. J. Leukoc. Biol. 74, 448–455 (2003). Sci. USA 107, 6204–6209 (2010).
51. Brouwer, C. P. J. M., Bogaards, S. J. P., Wulferink, M., Velders, M. P. & 71. Agrawal, S. et al. Riboswitch-mediated inducible expression of an
Welling, M. M. Synthetic peptides derived from human anti- astaxanthin biosynthetic operon in plastids. Plant Physiol. 188,
microbial peptide ubiquicidin accumulate at sites of infections and 637–652 (2022).
eradicate (multi-drug resistant) Staphylococcus aureus in mice. 72. Oey, M., Lohse, M., Scharff, L. B., Kreikemeyer, B. & Bock, R. Plastid
Peptides 27, 2585–2591 (2006). production of protein antibiotics against pneumonia via a new
52. Seo, J. K., Lee, M. J., Nam, B. H. & Park, N. G. CgMolluscidin, a novel strategy for high-level expression of antimicrobial proteins. Proc.
dibasic residue repeat rich antimicrobial peptide, purified from the Natl Acad. Sci. USA 106, 6579–6584 (2009).
gill of the Pacific oyster, Crassostrea gigas. Fish. Shellfish Immunol. 73. Elghabi, Z., Karcher, D., Zhou, F., Ruf, S. & Bock, R. Optimization of
35, 480–488 (2013). the expression of the HIV fusion inhibitor cyanovirin-N from the
53. Miyata, T. et al. Antimicrobial peptides, isolated from Horseshoe tobacco plastid genome. Plant Biotechnol. J. 9, 599–608 (2011).
Crab Hemocytes, Tachyplesin II, and Polyphemusins I and II: Che- 74. Li, S.-J. & Hochstrasser, M. A new protease required for cell-cycle
mical Structures and and biological activity. J. Biochem. 106, progression in yeast. Nature 398, 246–251 (1999).
663–668 (1989). 75. Hickey, C. M., Wilson, N. R. & Hochstrasser, M. Function and reg-
54. Pouny, Y., Rapaport, D., Shai, Y., Mor, A. & Nicolas, P. Interaction of ulation of SUMO proteases. Nat. Rev. Mol. Cell Biol. 13,
antimicrobial Dermaseptin and its fluorescently labeled analogs 755–766 (2012).

Nature Communications | (2022)13:5856 16

Article https://doi.org/10.1038/s41467-022-33516-1

76. Malakhov, M. P. et al. SUMO fusions and SUMO-specific protease culture, the MPI-MP GreenTeam for plant cultivation, the MPI-MP pho-
for efficient expression and purification of proteins. J. Struct. Funct. tographers for taking pictures of the plants, and Prof. Dr. Dirk Walther for
Genom. 5, 75–86 (2004). help with codon optimization of the linker sequences. This research
77. Marín-Medina, N., Ramírez, D. A., Trier, S. & Leidy, C. Mechanical project received funding from the Max Planck Society and the European
properties that influence antimicrobial peptide activity in lipid Research Council (ERC) under the European Union’s Horizon 2020
membranes. Appl. Microbiol. Biotechnol. 100, 10251–10263 (2016). research and innovation programme (ERC-ADG-2014; grant agreement
78. Zhang, Q.-Y. et al. Antimicrobial peptides: mechanism of action, No 669982) to R.B.
activity and clinical potential. Mil. Med. Res. 8, 48 (2021).
79. Wu, Y., Jayasekara Arachchige, M. N., Pei, X. & Wang, D. Effect of Author contributions
WAM-1 transgenic Arabidopsis thaliana on the development and M.P.H., J.F. and R.B. designed the experiments, M.P.H. performed the
gut microbial composition of diamondback moth (Plutella xylos- majority of the experiments, S.C. contributed to the development and
tella) larvae (Lepidoptera: Plutellidae). J. Appl. Entomol. 143, characterization of the SUMO tag system, C.K. and Z.T. contributed to
1011–1021 (2019). the molecular characterization of transplastomic plants, F.V.L. per-
80. Broekaert, W. F. et al. Antimicrobial peptides from plants. Crc. Crit. formed northern blots depicted in this article, S.A. and D.K. developed
Rev. Plant Sci. 16, 297–323 (1997). the split transformation system with RT7RP and PT7, D.K. also created
81. Flores, T., Alape-Girón, A., Flores-Dıaz,́ M., Flores, H. E. & Ocatin A expression vector pDK323, F.M. contributed to DNA construct genera-
novel tuber storage protein from the andean tuber crop oca with tion and development of protein purification methods, X.K. performed a
antibacterial and antifungal activities. Plant Physiol. 128, major part of the tissue culture and plant transformation experiments, all
1291–1302 (2002). authors interpreted and analyzed data, R.B. and M.P.H. wrote the paper
82. Campos, M. L., de Souza, C. M., de Oliveira, K. B. S., Dias, S. C. & with input from all other authors.
Franco, O. L. The role of antimicrobial peptides in plant immunity. J.
Exp. Bot. 69, 4997–5011 (2018). Funding
83. Gray, B. N., Ahner, B. A. & Hanson, M. R. Extensive homologous Open Access funding enabled and organized by Projekt DEAL.
recombination between introduced and native regulatory plastid
DNA elements in transplastomic plants. Transgenic Res. 18, Competing interests
559–572 (2009). The authors declare no competing interests.
84. Huang, F.-C. et al. Efficient plastid transformation in tobacco using
the aphA-6 gene and kanamycin selection. Mol. Genet. Genomics Additional information
268, 19–27 (2002). Supplementary information The online version contains
85. Apel, W., Schulze, W. X. & Bock, R. Identification of protein stability supplementary material available at
determinants in chloroplasts. Plant J. 63, 636–650 (2010). https://doi.org/10.1038/s41467-022-33516-1.
86. Chiu, W. et al. Engineered GFP as a vital reporter in plants. Curr. Biol.
6, 325–330 (1996). Correspondence and requests for materials should be addressed to
87. Murashige, T. & Skoog, F. A revised medium for rapid growth and Ralph Bock.
bio assays with tobacco tissue cultures. Physiol. Plant. 15,
473–497 (1962). Peer review information Nature Communications thanks Anil Day, Jörg
88. Doyle, J. J. & Doyle, J. L. A rapid total DNA preparation procedure for Durner, Ivo Frébort and the other, anonymous, reviewer(s) for their
fresh plant tissue. Focus. 12, 13–15 (1990). contribution to the peer review of this work. Peer reviewer reports are
89. Cahoon, E. B., Shanklin, J. & Ohlrogge, J. B. Expression of a available.
coriander desaturase results in petroselinic acid production in
transgenic tobacco. Proc. Natl Acad. Sci. USA 89, Reprints and permission information is available at
11184–11188 (1992). http://www.nature.com/reprints
90. Shäger, H. & von Jagow, G. Tricine-sodium dodecyl sulfate-
polyacrylamide gel electrophoresis for the separation of proteins in Publisher’s note Springer Nature remains neutral with regard to jur-
the range from 1 to 100 kDa. Anal. Biochem. 166, 368–379 (1987). isdictional claims in published maps and institutional affiliations.
91. Schägger, H. Tricine – SDS-PAGE. Nat. Protoc. 1, 16–23 (2006).
92. Steinberg, D. A. & Lehrer, R. I. Designer assays for antimicrobial Open Access This article is licensed under a Creative Commons
peptides. in antibacterial peptide protocols (ed. Shafer, W. M.) Attribution 4.0 International License, which permits use, sharing,
169–186 (Humana Press). https://doi.org/10.1385/0-89603-408-9: adaptation, distribution and reproduction in any medium or format, as
169 (1997). long as you give appropriate credit to the original author(s) and the
93. Lehrer, R. I., Rosenman, M., Harwig, S. S. S. L., Jackson, R. & source, provide a link to the Creative Commons license, and indicate if
Eisenhauer, P. Ultrasensitive assays for endogenous antimicrobial changes were made. The images or other third party material in this
polypeptides. J. Immunol. Methods 137, 167–173 (1991). article are included in the article’s Creative Commons license, unless
94. Fernandes, J. M. O., Molle, G., Kemp, G. D. & Smith, V. J. Isolation indicated otherwise in a credit line to the material. If material is not
and characterisation of oncorhyncin II, a histone H1-derived anti- included in the article’s Creative Commons license and your intended
microbial peptide from skin secretions of rainbow trout, Oncor- use is not permitted by statutory regulation or exceeds the permitted
hynchus mykiss. Dev. Comp. Immunol. 28, 127–138 (2004). use, you will need to obtain permission directly from the copyright
holder. To view a copy of this license, visit http://creativecommons.org/
Acknowledgements licenses/by/4.0/.
We thank Dr. Stephanie Ruf and her team (in particular Jennifer Ber-
gander and Luisa Heinig) for help with plant transformation and tissue © The Author(s) 2022

Nature Communications | (2022)13:5856 17