Differential Labeling of Chemically Modified Peptides and Lipids among Cyanobacteria Planktothrix and Microcystis

The cyanoHAB forming cyanobacteria Microcystis and Planktothrix frequently produce high intracellular amounts of microcystins (MCs) or anabaenopeptins (APs). In this study, chemically modified MCs and APs have been localized on a subcellular level in Microcystis and Planktothrix applying copper-catalyzed alkyne-azide cycloaddition (CuACC). For this purpose, three different non-natural amino acids carrying alkyne or azide moieties were fed to individual P. agardhii strains No371/1 and CYA126/8 as well as to M. aeruginosa strain Hofbauer showing promiscuous incorporation of various amino acid substrates during non-ribosomal peptide synthesis (NRPS). Moreover, CYA126/8 peptide knock-out mutants and non-toxic strain Synechocystis PCC6803 were processed under identical conditions. Simultaneous labeling of modified peptides with ALEXA405 and ALEXA488 and lipid staining with BODIPY 505/515 were performed to investigate the intracellular location of the modified peptides. Pearson correlation coefficients (PCC) obtained from confocal images were calculated between the different fluorophores and the natural autofluorescence (AF), and between labeled modified peptides and dyed lipids to investigate the spatial overlap between peptides and the photosynthetic complex, and between peptides and lipids. Overall, labeling of modified MCs (M. aeruginosa) and APs (P. agardhii) using both fluorophores revealed increased intensity in MC/AP producing strains. For Synechocystis lacking NRPS, no labeling using either ALEXA405 or ALEXA488 was observed. Lipid staining in M. aeruginosa and Synechocystis was intense while in Planktothrix it was more variable. When compared with AF, both modified peptides and lipids showed a heterologous distribution. In comparison, the correlation between stained lipids and labeled peptides was not increased suggesting a reduced spatial overlap.


Introduction
Planktonic toxin-producing cyanobacteria of the genera Microcystis and Planktothrix frequently form algal blooms in freshwater systems. The accumulation of cyanobacterial biomass due to cyanobacterial harmful algal blooms (CHABs) and their toxic or bioactive metabolites can cause diseases or even death of animals drinking the polluted waters. This can also be harmful to people if the blooms are produced in drinking water sources, and it may be due to their prolific secondary metabolism why cyanobacteria increase particularly in habitats influenced by eutrophication.
The microcystins (MCs) and anabaenopeptins (APs) are among the most common peptides produced by cyanobacteria. These two peptide families are synthesized via non-ribosomal peptide synthesis (NRPS) by large multifunctional enzyme complexes, using partly non-proteinogenic amino acids as substrates [1]. MCs are well known for their toxicity as inhibitors of protein-phosphatase 1 and 2A in the nanomolar range. APs Azide and alkyne groups do not occur endogenously and therefore are less likely to interact with other molecules inside the cytoplasm [13,14]. Natural mutations affecting certain adenylation domains of the respective NRPS pathway lead to a more promiscuous activation of the substrate allowing the incorporation of non-natural amino acids into the synthesized MCs and APs ( Figure 1) [4,15,16] enabling targeting those molecules, i.e., with azide and alkyne carrying fluorophores through a "copper catalyzed azide-alkyne cycloaddition" (CuACC) [12,13].

Study Organisms
P. agardhii strain No371/1 (isolated from Moose Lake, Alberta, Canada by Rainer Kurmayer in 2005) and P. agardhii (NIVA-)CYA126/8 (isolated from Lake Langsjön, Stockholm, Sweden by Olav Skulberg in 1984) are known to carry a promiscuous adenylation domain ApnAA1, leading to the production of structural AP variants with variable amino acids occurring in the exocyclic position 1 [2]. P. agardhii CYA126/8 mutants with experimentally inactivated AP synthesis (∆apnC), cyanopeptolin synthesis (∆ociA), microviridin synthesis (∆mvdC), and microcystin synthesis (∆mcyD) were included. Those knock out mutants have been generated earlier during the elucidation of the respective peptide synthesis pathways [4,[17][18][19][20]. In order to reduce genetic variability, strains were regrown from one filament according to the standard filament isolation technique on agar [21]. Moreover, the MC-producing Microcystis aeruginosa strain Hofbauer (isolated from Lake Neusiedl, Burgenland, Niederösterreich, Austria by Barbara Hofbauer in 1982) was included for labeling MC structural variants. Finally, the strain Synechocystis PCC6803, which lacks NRPS coding genes [22], was included to test the specificity of the non-natural amino acid incorporation in general.

Growth Conditions
The strains were cultured in BG11 medium [21] at 20 • C and 40-60 µE m −2 s −1 semi-continuously following the turbidostat principle [23]. For the four gene inactivation mutants 1 µg mL −1 of chloramphenicol was added as a selection marker. During experimental feeding of non-natural amino acids, no further chloramphenicol was added. Precultures were started from cultures maintained at maximum growth rate conditions until they reached an optical density (OD 600nm ) of 0.1 (1 cm light path), then diluted to an OD 600nm of 0.01 and supplemented once either with 50 µM L-4-azidophenylalanine (Phe-Az), (Carl Roth, Karlsruhe, Germany), N-propargyloxy-carbonyl-L-lysine (Prop-Lys), (Sichem, Bremen, Germany) or O-propargyl-L-Tyrosine (Prop-Tyr) (Iris Biotech, Marktredwitz, Germany) dissolved in 1 mM NaOH. Control cultures were supplied with 1 mM NaOH only. Cells were finally collected after reaching an OD 600nm 0.1 (which was after 6 days for M. aeruginosa and P. agardhii and after 3 days for Synechocystis PCC6803). Three culture flasks per strain were inoculated under identical conditions, which were intended for the measuring of OD 600nm , the labeling of the bioactive peptides and the peptide analysis through HPLC-MS ( Figure S1).
In general, cells were sedimented by centrifugation (14,000× g), washed with fresh PBS (3×), and fixed with 2% PFA (15 min). Samples were then washed again with PBS (3×) and permeated by incubating them for 10 min in PBS containing Triton X-100 (0.1%, v/v) and washed with PBS (3×). The resulting pellets were resuspended in 50 µL fresh PBS and stored at 4 • C.
For click-chemistry labeling, all cells were processed under identical conditions, the lipids were labeled through modified Brennan's protocol [25,26]. In brief, cells were incubated in PBS at 37 • C for 10 min followed by a 10 min incubation with 4 µM BODIPY 505/515 at RT in the dark and washing excess of the fluorophore with PBS. Consecutively, the modified bioactive peptides were labeled using copper-catalyzed azide-alkyne cycloaddition (CuAAC) click reaction chemistry [13]. The cells were washed with 2% bovine serum albumin (BSA) while the reaction was performed according to manufactures instructions (Thermo Fisher Scientific Click-It ® Cell Reaction Buffer Kit), i.e., in a Trisbuffered reaction mix with 1 mM CuSO 4 and various additives [13]. Supplemented blue fluorophore ALEXA405-azide or green fluorophore ALEXA488-azide were used to label modified peptides carrying alkyne moieties Prop-Tyr and Prop-Lys. Vice versa modified peptides carrying Phe-Az were detected via ALEXA488-alkyne. The concentrations at which the fluorophores were supplemented into the labeling reaction mix were 40 µM and 4 µM for ALEXA405-azide and ALEXA488-azide/alkyne, respectively. The reaction mix was incubated for 1 h in the dark and washed with BSA 2% and resuspended in 50 µL fresh PBS [12].
All strains were grown in the absence and presence of the three non-natural amino acids (Phe-Az, Prop-Lys and Prop-Tyr). Controls were made from cells grown in the absence of amino acids but treated under identical conditions. In addition, all treatments were processed and inspected under the confocal microscope without fluorophore addition. Finally, 5 µL of the samples were mounted using antifade solution (ProLong ® Diamond Antifade Mountant, Thermo Fisher Scientific, Darmstadt, Germany). The samples were air-dried for 48 h in the dark and stored at 4 • C until microscopical analysis.

Peptide Extraction
In parallel to cell preparation for labeling, cells were harvested via filtration using pre-weighed glass fiber (GF/C) filters. The filters with the collected biomass were dried using a vacuum centrifuge at RT for 4 h. The peptide content from the strains was extracted from the dried biomass using aqueous methanol (50% v/v) as described previously [27].

HPLC-MS Analysis
MCs and APs were separated by HPLC (HP 1100, Agilent, Vienna, Austria) system, using a linear mobile phase of water/acetonitrile (0.05% trifluoroacetic acid) gradient from 80:20 to 50:50 in 45 min at a flow rate of 1 mL/min through a LiChroCART 250-4 cartridge system (Merck, Darmstadt, Germany) with LiChrospher 100 octyldecyl silane (ODS), (5 µm particle size) as the solid reversed phase [12]. The HPLC system was coupled to an ESI-MS (Electrospray Ionization Mass Spectrometer) ion trap (amaZon SL Ion Trap MS, Bruker Daltonik, Bremen, Germany) operated in positive ionization mode. A mixture of nitrogen and helium was used as sheath gas and collision gas, respectively (43 psi, 9 L/min, 250 • C) with 5 kV capillary voltage. MC and AP variants were assigned according to their retention time, protonated mass [M+H] + , and fragmentation patterns. Fragmentation was achieved by automated fragmentation and adjusted to MS 2 for the two most abundant molecules showing the highest intensity while for MS 3 only the peak with maximum intensity was further fragmented. LC-MS chromatograms and fragmentation patterns were investigated using the Bruker Compass data analysis software (version 4.2), (Bruker Daltonik, Bremen, Germany). Under these conditions the limit of detection for analytical standards MC-RR [M+H] + 1038.5, MC-YR [M+H] + 1045.5, and MC-LR [M+H] + 995.5 was 10 ng injected (Cyanobiotech, Berlin, Germany).

Microscopic Analysis
Confocal images were acquired on a SP8 laser scanning microscope (Leica Microsystems, Wetzlar, Germany) at the Biooptics facilites (CCB) from Medizinische Universität Innsbruck. Images were acquired at an XY resolution of 50 nm and Z resolution of 150 nm. A total of 20 individual cells (Microcystis or Synechocystis) or filaments (Planktothrix) were randomly selected per strain and treatment. Confocal images were deconvolved to improve final resolution with Huygens Essential 20.04 software using the Classic Maximum Likelihood Estimation (CMLE) algorithm according to the manufacturer's recommendations (Science Volume Imaging BV, Hilversum, The Netherlands).
Finally, total intensities were obtained and compiled for the three RGB channels, as well as the ratios between the different channels. Specifically, the green channel was used to measure the labeling with ALEXA488 and BODIPY 505/515, the blue channel measured the labeling effect with ALEXA405, while the red channel was used to measure the autofluorescence (AF) of the cells (Figure 2).

Microscopic Analysis
Confocal images were acquired on a SP8 laser scanning microscope (Leica Microsystems, Wetzlar, Germany) at the Biooptics facilites (CCB) from Medizinische Universität Innsbruck. Images were acquired at an XY resolution of 50 nm and Z resolution of 150 nm. A total of 20 individual cells (Microcystis or Synechocystis) or filaments (Planktothrix) were randomly selected per strain and treatment. Confocal images were deconvolved to improve final resolution with Huygens Essential 20.04 software using the Classic Maximum Likelihood Estimation (CMLE) algorithm according to the manufacturer's recommendations (Science Volume Imaging BV, Hilversum, The Netherlands).
Finally, total intensities were obtained and compiled for the three RGB channels, as well as the ratios between the different channels. Specifically, the green channel was used to measure the labeling with ALEXA488 and BODIPY 505/515, the blue channel measured the labeling effect with ALEXA405, while the red channel was used to measure the autofluorescence (AF) of the cells (Figure 2).

Colocalization Coefficient
Calculation of the colocalization coefficients between the differently labeled molecules was determined using Huygens Essential 20.04 software built in Huygens Colocalization Analyzer Advanced (Scientific Volume Imaging BV, Hilversum, The Netherlands). The background of the images was corrected using the integrated Costes method for background estimation, by calculating a regression line in which each point on the line is a combination of backgrounds in both channels and estimates the position where the Pearson coefficient of the background is zero [28].
Overall, the Pearson colocalization coefficients (PCC) measures the three dimensional voxel intensity covariance between the signal in two different channels [29]. PCC values range from −1 meaning a perfectly opposed distribution in the signal intensities and 1, which represents a complete overlap of the signals. The deconvolved images were analyzed using pairwise comparisons between the blue channel (ALEXA405) and the red channel (AF), thus calculating the PCC between the peptide intensity labeled with ALEXA405 and the natural AF. The intensities measured from the green channel (either BODIPY 505/515 or ALEXA488) were compared against the red channel (AF) for the colocalization between the lipids dyed with BODIPY 505/515 and the AF, or the intensity of ALEXA488 labeled peptides and AF. Finally, the blue channel (ALEXA405) and the green channel (BODIPY 505/515) were processed to determine the PCC between the labeled peptides and the labeled lipids.

Growth Rate
In general, the supplementation of the medium using the non-natural amino acids Prop-Tyr and Prop-Lys did not reduce the growth rate of any of the strains, i.e., the growth rates varied between 0.27-0.45 d −1 for Planktothrix strains, while for cultures without non-natural amino acids the growth rates ranged between 0.32-0.47 d −1 . In general, the unicellular cyanobacteria M. aeruginosa and Synechocystis PCC6803 showed higher growth rates ranging from 0.55-0.57 d −1 and 0.96-1.01 d −1 (Table S1).
In contrast to Prop-Tyr and Prop-Lys the addition of Phe-Az resulted in a decline in growth rate among all the strains. In particular, the growth rate of Synechocystis PCC6803 and P. agardhii CYA126/8 ∆apnC declined the most from Phe-Az addition, while P. agardhii CYA126/8 WT was less affected. Filaments from cultures grown in the presence of Phe-Az frequently showed a reduced AF, suggesting a reduced vitality of the cells.
Accordingly, obtained dry weights were smaller for cells harvested from cultures grown in the presence of Phe-Az when compared with Prop-Tyr or Prop-Lys treatments. Consequently, for Phe-Az relatively low amounts of biomass were available for peptide extraction (0.5-1.4 mg of dry weight) whereas for Prop-Tyr and Prop-Lys dry weights ranged from 0.9-3.7 mg and 1.5-4.0 mg, respectively.

Modified Peptides
As compared to control cells grown in the absence of non-natural amino acids we detected chemically modified MCs in M. aeruginosa peptide extracts carrying incorporated azide or alkyne moieties ( Figure S5 (Table 1). For Synechocystis PCC6803 the observed elution profile was not assigned to respective compounds, however, compared with M. aeruginosa no change in chromatogram elution profile was observed through the addition of non-natural AA ( Figure S6). Modified AP variants carrying alkyne or azide moieties were detected in P. agardhii No371/1 carrying promiscuous ApnAA 1 domains ( Figure S7). Specifically, we were able to detect the incorporation of Phe-Az into the AP molecule presumably in its reduced form AP-Phe-azide [ Figure S8). Any AP structural variant was observed for the ∆apnC mutant ( Figure S9). We were not able to detect AP incorporating Prop-Lys for CYA126/8 WT and its ∆ociA, ∆mvdC, ∆mcyD peptide knock out mutants (Figures S10-S12

Peptide Labeling Intensity
Both ALEXA488 and ALEXA405 fluorophores resulted in peptide labeling either in M. aeruginosa or in P. agardhii. For all non-natural amino acids (Phe-Az, Prop-Lys and Prop-Tyr) labeling with ALEXA488 fluorophore was observed, however, labeling in the non-toxic cyanobacteria Synechocystis PCC6803 was not detected. In general, the green fluorophore ALEXA488 produced brighter signals with the highest intensities and a more pronounced difference from background fluorescence ( Table 2). The blue fluorophore ALEXA405 also resulted in peptide labeling, however, the intensity was generally lower and the distinction from background fluorescence such as natural AF was reduced ( Table 3).
The toxic cyanobacteria M. aeruginosa showed a significant increase in labeling intensities using all three amino acids when compared with control cells grown in the absence of any amino acid but treated under identical conditions. In particular, Phe-Az, Prop-Lys and Prop-Tyr resulted in five-and four-fold increases in the average intensity compared to the control (Table 2). Correspondingly, strong labeling was detected for Prop-Lys and Prop-Tyr fed cultures when ALEXA405 was applied, but the increase was only 1-2.4 fold. Since ALEXA405 was only available as an azide it could not be applied for MC or AP labeling carrying the Phe-Az moiety (Table 3). In contrast, Synechocystis PCC6803 did not show any intensity increase with any of the amino acids fed and the application of ALEXA405.
Among the Planktothrix strains significant increase in intensity using the green ALEXA488azide was consistently detected for No371/1. For Phe-Az, however, fluorescence intensity was not significantly increased using strain No371/1 (0.6 ± 0.7 vs. 1.3 ± 0.6, without and with ALEXA488-alkyne respectively). Results were less consistent using the blue fluorophore ALEXA405: For strain No371/1 only Prop-Lys resulted in a significant increase of fluorescence intensity (1.2 ± 0.2 vs. 0.8 ± 0.2 with and without ALEXA405-azide respectively, p < 0.001), while no evidence of labeling was found in the cultures fed with Prop-Tyr (0.9 ± 0.3 vs. 1.0 ± 0.2). Table 2. Average (±SD) min-max green fluorescence intensity obtained for individual treatments using non-natural amino acid feeding (Phe-Az, Prop-Lys, and Prop-Tyr) and subsequent labeling by ALEXA488 using copper-catalyzed azid-alkyne cycloaddition (CuAAC). The intensity was divided by the average intensity of control filaments or cells, i.e., cells which were grown without amino acid addition but used for the chemical reaction under identical conditions. No Fluorophore indicates filaments or cells grown with amino acid addition but no subsequent labeling by the click-chemical reaction. n: number of individual filaments (Planktothrix) or cells (Microcystis, Synechocystis).

ALEXA405
The cyanopeptolin synthesis mutant CYA126/8 ∆ociA showed significantly increased green fluorescence intensity for all the non-natural amino acids supplements. In comparison with the CYA126/8 WT the ∆ociA mutant showed higher fluorescence intensity, i.e., 1. Similar results were obtained for microviridin (∆mvdC) and MC (∆mcyD) gene knock out mutants both showing increased intensity of green fluorescence using ALEXA488, i.e., 1.7-fold vs. 0.6-0.9 fold for Prop-Lys and 1.5-1.6-fold vs. 0.7-0.9-fold for Prop-Tyr. Significant ALEXA488 labeling using Phe-Az was observed for the ∆mvdC mutant only. ALEXA405 labeling again was overall reduced and was found increased for the ∆mvdC and the ∆mcyD strain fed with Prop-Lys (Table 3).

Peptide Intensity/Autofluorescence Ratio
Both genera M. aeruginosa and P. agardhii increased in the ratio between ALEXA488 and AF with either of the three non-natural amino acids. Whereas Synechocystis PCC6803 did not show this effect. According to the increased green intensity (Table 2), the highest signal ratios were observed for M. aeruginosa, i.e., a median ratio of 1.5 for Phe-Az, 0.9 for Prop-Lys, and 1.2 for Prop-Tyr as compared with 0.2 for controls ( Figure 3).   . Controls were grown without amino acid addition but used for the chemical reaction under identical conditions. No Fluorophore indicates filaments or cells grown with amino acid addition but no subsequent labeling by click-chemical reaction. The gradient in coloring was defined for each strain separately using the average intensity from the control cultures. Superscripts indicate statistically significant different subgroups after overall difference was found (p < 0.05).
Among Planktothrix strains, P. agardhii No371/1 produced the highest ratios, i.e., 1.0 for Phe-Az, 1.2 for Prop-Lys, and 0.7 for Prop-Tyr when compared with 0.3 for controls. For CYA126/8 WT the ratios using the feeding treatments Prop-Lys and Prop-Tyr remained unaltered, while Phe-Az (0.2) increased when compared with controls (0.1). The ∆apnC mutant did not show increased ratios of peptide intensity vs. AF between the treatments.
The mutants ∆ociA, ∆mvdC and ∆mcyD showed rather similar response, i.e., ratios increased for Phe-Az (0.3-0.6), Prop-Lys (0.3-0.4), Prop-Tyr (0.3-0.4) vs. 0.2 for controls. ALEXA405 ratios were calculated by dividing the blue labeling intensity by AF (Figure 4). In general, both ALEXA488 and ALEXA405 signal ratios showed comparable increases, whereas Synechocystis PCC6803 did not show response compared to the control. coloring was defined for each strain separately using the average intensity from the control cultures. Superscripts indicate statistically significant different subgroups after overall difference was found (p < 0.05). . Controls were grown without amino acid addition but used for the chemical reaction under identical conditions. No Fluorophore indicates filaments or cells grown with amino acid addition but no subsequent labeling by click-chemical reaction. The gradient in coloring was defined for each strain separately using the average intensity from the control cultures. Superscripts indicate statistically significant different subgroups after overall difference was found (p < 0.05).

Change in Supplementary Materials:
The Supplementary Materials were changed accordingly and were included as a separate document.
The authors would like to apologize for any inconvenience caused to the readers by these changes. . Controls were grown without amino acid addition but used for the chemical reaction under identical conditions. No Fluorophore indicates filaments or cells grown with amino acid addition but no subsequent labeling by click-chemical reaction. The gradient in coloring was defined for each strain separately using the average intensity from the control cultures. Superscripts indicate statistically significant different subgroups after overall difference was found (p < 0.05).
In contrast, among P. agardhii strains, No371/1 the BODIPY signal was rather low, i.e., 0.4-0.6 vs. 0.3-0.6 in controls. After any increase in fluorescence intensity, the median ratio for CYA126/8 WT also did not differ from treatments without labeling, i.e., 0.1-0.1 vs. 0.1-0.2 (controls). Similarly, for the ∆apnC mutant, no significant change in green intensity was observed. In contrast, median ratios increased significantly for mutants ∆ociA

Colocalization of Peptides vs. Autofluorescence and Lipids
From the same deconvolved images used for comparing signal intensities, the Pearson colocalization coefficient (PCC) was calculated pairwise between blue, green, and red fluorescence signals ( Figure 6). The ALEXA405 labeled peptide signal disposition relative to the AF and lipids were measured with the voxel comparison of the total intensity in the blue channel (A405) against the red (AF) and green (BODIPY 505/515) channels. The same procedure was followed for determining the lipid (green channel) colocalization coefficient against the AF (red channel) and the ALEXA488 labeled peptide distribution (green channel) against the AF (red channel). Since significant BODIPY/peptide labeling only occurred in M. aeruginosa, P. agardhii No371/1, CYA126/8 ∆ociA and CYA126/8 ∆mcyD only these strains were used for colocalization analysis.
For lipids, the PCC values with the AF were often more variable and in general increased i.e., M. aeruginosa PCC values of 0.52 ± 0.18 were found, for P. agardhii No371/1 0.86 ± 0.15, and for CYA126/8 mutants ∆ociA 0.70 ± 0.15 and ∆mcyD 0.70 ± 0.12. Thus, regarding autofluorescence (AF) both ALEXA488 labeled peptides and BODIPY stained lipids showed a more heterogeneous distribution in the cell. When compared with ALEXA488 for most treatments, the PCC between lipids and AF was found to have significantly increased.
In contrast to ALEXA488 labeled peptides, the ALEXA405 labeled peptides showed higher and less variable correlation with AF, i.e., for M. aeruginosa PCC values ranged from 0.65 ± 0.05, for No371/1 0.80 ± 0.04, for ∆ociA 0.83 ± 0.03 and for ∆mcyD 0.78 ± 0.03. Furthermore, the correlation between ALEXA405 labeled peptides and BODIPY 505/515 stained lipids was found to be overall variable and reduced, i.e., for M. aeruginosa calculated values were 0.44 ± 0.19, for No371/1 between 0.72 ± 0.17, and for CYA126/8 mutants ∆ociA and ∆mcyD PCC values ranged between 0.72 ± 0.14 and 0.65 ± 0.15 respectively. In summary, when compared with AF, the spatial correlation between ALEXA405 labeled peptides and lipids was found rather reduced than increased.  . Controls were grown without amino acid addition. No Fluorophore indicates filaments or cells grown with amino acid addition but no BODIPY labeling. The gradient in coloring was defined using the average intensity from the control cultures without supplementing amino acids. Superscripts indicate statistically significant different subgroups after overall difference was found (p < 0.05); n/d: no data.
In contrast, among P. agardhii strains, No371/1 the BODIPY signal was rather low, i.e., 0.4-0.6 vs. 0.3-0.6 in controls. After any increase in fluorescence intensity, the median ratio for CYA126/8 WT also did not differ from treatments without labeling, i.e., 0.1-0.1 vs. 0.1-0.2 (controls). Similarly, for the ΔapnC mutant, no significant change in green intensity was observed. In contrast, median ratios increased significantly for mutants ΔociA  . Controls were grown without amino acid addition. No Fluorophore indicates filaments or cells grown with amino acid addition but no BODIPY labeling. The gradient in coloring was defined using the average intensity from the control cultures without supplementing amino acids. Superscripts indicate statistically significant different subgroups after overall difference was found (p < 0.05); n/d: no data. procedure was followed for determining the lipid (green channel) colocalization coefficient against the AF (red channel) and the ALEXA488 labeled peptide distribution (green channel) against the AF (red channel). Since significant BODIPY/peptide labeling only occurred in M. aeruginosa, P. agardhii No371/1, CYA126/8 ΔociA and CYA126/8 ΔmcyD only these strains were used for colocalization analysis. (blue channel) and BODIPY 505/515 (green channel). Pairwise statistical analysis between the PCC values were calculated i.e., ALEXA488 labeled peptides and natural autofluorescence (AF), BODIPY 505/515 stained lipids and AF, ALEXA405 labeled peptides and natural AF and ALEXA405 labeled peptides, and BODIPY 505/515 stained lipids. Superscripts indicate statistically significant different subgroups after overall difference was found (p < 0.05); n/d: no data; n/a: not applicable.
For lipids, the PCC values with the AF were often more variable and in general increased i.e., M. aeruginosa PCC values of 0.52 ± 0.18 were found, for P. agardhii No371/1 0.86 ± 0.15, and for CYA126/8 mutants ΔociA 0.70 ± 0.15 and ΔmcyD 0.70 ± 0.12. Thus, regarding autofluorescence (AF) both ALEXA488 labeled peptides and BODIPY stained lipids showed a more heterogeneous distribution in the cell. When compared with ALEXA488 for most treatments, the PCC between lipids and AF was found to have significantly increased.

Discussions
In general, the supplementation of the non-natural amino acids Prop-Lys and Prop-Tyr did not produce a detectable adverse effect in the growth of the strains in this study. In contrast, Phe-Az feeding resulted in a decrease in the growth rate for many strains (Table S1). The inhibitory effects of sodium azide (NaN 3 ) are well documented [34][35][36]. Sodium azide inhibits the cytochrome oxidase in bacteria. Similar to NaN 3 the azide ion from Phe-Az may block the catalytic domain of cytochrome-c-oxidase binding oxygen.
When comparing LC-MS elution profiles ( Figures S5-S12), a rather specific incorporation of the non-natural amino acids into the MCs or APs produced by NRPS synthesis has been observed, i.e., using strains carrying promiscuous A-domains such as McyBA1 of M. aeruginosa and Planktothrix agardhii ApnAA 1 in strains No371/1 and CYA126/8 [2]. Thereby, newly modified MC/AP structures, presumably carrying the Phe-Az moiety either in pos.2 of the MC molecule or in the exocyclic position of the AP molecule were obtained. In addition, no labeling in the non-toxic strain Synechocystis PCC6803 was detected, demonstrating that unspecific incorporation of non-natural amino acids Phe-Az, Prop-Lys and Prop-Tyr into anabolic pathways other than NRPS did not occur. Nevertheless, the ∆apnC mutant revealed Prop-Tyr labeling, possibly suggesting ongoing partial AP synthesis. The AP synthesis pathway through NRPS is starting from ApnAA1 with activating and condensing the first exocyclic amino acid to the conserved lysine via the characteristic ureido-linkage [4]. Thus even in ∆apnC mutant, the AP synthesis may continue with a non-natural AA through the presumably intact NRPS ApnA2 and ApnB [4], and a linear AP peptide fragment comprising the three amino acids Tyr/Arg-Lys-Val might still be synthesized.
Currently the observed, rather moderate, Prop-Tyr labeling for the P. agardhii ∆apnC mutant strain does not support our hypothesis on Prop-Tyr incorporation during MC biosynthesis, i.e., identification of its derivation from D-Asp-MC-Tyr-alkyne [M + H] 1069.5 could not be unequivocally performed. However, since neither the ALEXA488 nor the ALEXA405 fluorescence to AF ratios were affected, we conclude that the signal increase from Tyr-alkyne in the ∆apnC mutant strain was generally minor.
Nonetheless, the incorporation of Prop-Lys could not be shown for AP synthesis using strain CYA126/8 WT or its mutants ∆ociA, ∆mvdC, ∆mcyD. It is possible, that low proportion of putative AP incorporating Prop-Lys is coeluting with natural AP 915, which occur with more than 50% proportion of total APs in strain CYA126/8. Since cyanobacteria are in general well known for their ability to uptake various AA from the medium [37], the labeling of non-natural AA transported across the inner cell membrane potentially creates artificial results. In this study for nearly all strains including Synechocystis PCC6803 low-albeit-significant amounts of non-natural AA (Phe-Az, Prop-Lys and Prop-Tyr) were detected during inspection of LC-MS chromatograms ( Figures S5-S12). Since Synechocystis PCC6803 also contained non-natural AA ( Figure S6) but did not show increase in ALEXA488 or ALEXA405 signal intensity (Figures 3 and 4), this potential artifact is considered less likely.
Using the ALEXA488 fluorophore the CYA126/8 WT strain showed overall lower intensity when compared to the peptide knock-out mutants ∆ociA, ∆mvdC, ∆mcyD as well as strain No371/1 (Figures 3 and 4). It has been suggested earlier that metabolite synthesis between different NRPS pathways might interfere and might depend on the availability of building blocks, precursors, and other resources. For example, the feeding of the putatively limiting amino acid D-L-homotyrosine led to the increased production and subsequent structural elucidation of AP 908 and 915 vs. cyanopeptolin 880 and 960 for strain CYA126/8 WT [33]. Thus, for unknown reasons the inactivation of cyanopeptolin, microviridin, or MC synthesis might favor the integration of unnatural AA into AP exocyclic position No1. Furthermore, the ApnAA1 domain of strain No371/1 has been found more promiscuous when compared with strain CYA126/8 WT [2], i.e., AP C carrying lysine in pos.1 is produced in low amounts by strain No371/1 but not by strain CYA126/8. Thus, strain No371/1 might be more efficient in the integration of the three tested non-natural AA into the exocyclic position of the AP molecule (Phe-Az, Prop-Lys Prop-Tyr).
Labeling with the blue fluorophore ALEXA405-azide yielded acceptable fluorescence signals for M. aeruginosa but lower signals for P. agardhii when grown in the presence of Prop-Lys and Prop-Tyr. In particular, peptide labeling results in P. agardhii CYA126/8 WT, and its peptide knock out mutants were more variable when compared to the data obtained with ALEXA488. One reason for the higher variability might be the optical interference of ALEXA405 with photosynthetic pigments. Indeed, AF recorded at 405 nm was found much higher than AF recorded at 490 nm ( Figure 2 vs. Figures S2-S4). Consequently, an increased background fluorescence for P. agardhii was found in the blue range for all three non-natural AA treatments, which possibly interferes with peptide labeling using ALEXA405 (Figure 3 vs. Figure 4). In general, the genus Planktothrix has shown higher AF when compared to Microcystis and Synechocystis, demonstrating an important ecophysiological trait in response to the metalimnetic and shade (low light)-adapted lifeform of Planktothrix in general. Technically, ALEXA405 was found useful for M. aeruginosa and the co-labeling of peptides vs. other intracellular organelles using green fluorescent dyes. However, for Planktothrix the stronger autofluorescence will possibly limit the applicability of ALEXA405 to strains with naturally lower AF.
Nevertheless, as a proof of concept, we differentially labeled lipids in parallel to labeled modified MC vs. AP variants and compared lipid labeling intensity between strains ( Figure 5). Both unicellular cyanobacteria revealed significant BODIPY labeling suggesting that our modified BODIPY protocol was functional. For Planktothrix, however, lipid labeling results appeared to be strain specific. While CYA126/8 WT strain and ∆apnC did not show any significant BODIPY signal, the three other mutants (∆ociA, ∆mvdC, ∆mcyD) showed higher intensity. Out of the three mutants showing increased BODIPY intensity, the ∆mvdC showed a decreased BODIPY/AF ratio, probably because of the increased AF.
Finally, only four strains were retained to quantify and compare the intracellular location of the peptides using the PCC between the ALEXA405 labeled peptide and the BODIPY 505/515 lipid signal. The distribution of modified peptides labeling signals with either ALEXA405 (blue channel) or ALEXA488 (green channel) against AF was also compared. As a result, compared with the correlation between the ALEXA405 (blue channel) and AF, the PCC values between the modified peptides (blue channel) and lipids (green channel) were found in a comparable upper range or decreasing. These results suggest that for M. aeruginosa and P. agardhii No371/1 both strains showed a relatively high spatial correlation between labeled peptides with both fluorophores against the natural AF ( Figure 6). Previous studies using cryosectioning also detected MCs alongside the thylakoidal membranes [11,38]. Other studies reported the majority of MCs alongside polyphosphate bodies [39].
On the other hand, P. agarhii CYA126/8 mutants ∆ociA and ∆mcyD showed an overall lower correlation between peptide ALEXA488 with AF but a similar range in PCC between lipids and AF distribution ( Figure 6). Indeed, from images, a more distinct pattern of labeled modified APs with ALEXA488 and ALEXA405, seemingly locating APs close to the septa separating different cells of the filament has been recognized. Whereas detection of modified APs alongside cellular septa is more evident for ALEXA488 (Figure 2), labeled peptides with ALEXA405 did not produce a reduction in PCC values probably due to the higher interference by AF (Figures S2-S4). Nonetheless, our results suggest that modified APs in P. agardhii ∆ociA and ∆mcyD may present a different intracellular disposition from M. aeruginosa or P. agardhii No371/1. It might be worthwhile to address the reproducibility of the intracellular location of APs in Planktothrix strains using the ALEXA488 labeling technique.
The protocols here described offer us a new tool to investigate the distribution of modified peptides inside the cell. More advanced image analysis, including 3D rendering would allow acquisition of additional measures such as volume of the detected modified MCs/APs, the number of individual vesicles, or distance from the membranes.
New developments and description of proteins associated with filamentous cyanobacteria, as the description of the cyanobacterial cell division protein CyDiv [40] may allow us to confirm the potential location of the labeled peptides alongside the cellular septa. This test would consist of the simultaneous application of the current CuAAC labeling technique and suitable staining of CyDiv protein followed by colocalization coefficient analysis.

Conclusions
In summary, there is little evidence that other nontarget peptides than APs or MCs are modified. Together with the positive labeling results for M. aeruginosa and the negative labeling results from Synechocystis PCC6803 and the CYA126/8 AP mutant we have reason to state that labeling is indeed related to AP/MC to a significant amount (and less to nontarget compounds). The generally increased AP labeling in the CYA126/8 mutants ∆ociA, ∆mvdC, ∆mcyD as compared to the WT can be because more resources are available for peptide synthesis since one peptide synthesis pathway is inactive. The results do not support the role of lipids spatially related to the intracellular accumulation of (chemically modified) MCs vs. APs. In other words, lipids were found less correlated with the MC/AP signal than the MC/AP signal with AF.