Nuclear and peroxisomal targeting of catalase

Abstract Catalase is a well‐known component of the cellular antioxidant network, but there have been conflicting conclusions reached regarding the nature of its peroxisome targeting signal. It has also been reported that catalase can be hijacked to the nucleus by effector proteins of plant pathogens. Using a physiologically relevant system where native untagged catalase variants are expressed in a cat2‐1 mutant background, the C terminal most 18 amino acids could be deleted without affecting activity, peroxisomal targeting or ability to complement multiple phenotypes of the cat2‐1 mutant. In contrast, converting the native C terminal tripeptide PSI to the canonical PTS1 sequence ARL resulted in lower catalase specific activity. Localisation experiments using split superfolder green fluorescent protein revealed that catalase can be targeted to the nucleus in the absence of any pathogen effectors, and that C terminal tagging in combination with alterations of the native C terminus can interfere with nuclear localisation. These findings provide fundamental new insights into catalase targeting and pave the way for exploration of the mechanism of catalase targeting to the nucleus and its role in non‐infected plants.

Targeting of peroxisomal matrix proteins utilises two types of peroxisomal targeting signals, PTS1 and PTS2, which are recognised by cytoplasmic receptors PEX5 and PEX7, respectively (Gould et al., 1989;Kato et al., 1996;Nito et al., 2002). Recently, a third type of PTS targeting signal has been identified which is known as PTS3.
Unlike the linear PTS1 and PTS2, the PTS3 is a signal patch (Kempiński et al., 2020). Non-PTS proteins are also able to import into peroxisomes by a mechanism called piggy-back import (Glover et al., 1994;Lee et al., 1997).
In general, the C-terminus tripeptide consensus sequence of PTS1 is (S/A/C)-(K/R/H)-L (Brocard & Hartig, 2006) which is obligatorily at the C terminus of the protein and binds within a funnelshaped binding pocket of the C terminal TPR domain of PEX5 (Gatto et al., 2003;Stanley et al., 2006) but the C-terminus of CAT2 (which is PSI) does not fit this consensus (reviewed in Mhamdi et al., 2012).
Several studies have addressed the peroxisomal targeting of plant CAT. For example, the last four amino acids of cottonseed CAT (Ccat) were required for targeting of a reporter protein to tobacco BY-2 suspension culture cells . In contrast, the last 10 amino acids of the pumpkin CAT were not required for the peroxisomal localisation of green fluorescent protein (GFP) in stably transformed BY-2 cells (Kamigaki et al., 2003). Other mutations elsewhere in the protein also prevented import (Fujikawa et al., 2019).
In this study, we have reinvestigated the requirements for targeting the A. thaliana CAT2 (AtCAT2). We identified an alternative splice variant of CAT2 gene in the TAIR database (At4g35090.2). This variant arises from non-splicing of the last intron, which removes the last 18 amino acids of the CAT protein. Using a combination of physiological, biochemical and cell biology approches, we analysed the effects of introducing AtCAT2 variants with modified C-termini into the cat2-1 mutant background. The comprehensive evaluation of activity, function and targeting in a physiologically relevant context revealed that the last 18 amino acids of the CAT2 sequence are dispensable for growth, cellular redox status and targeting of the enzyme to the peroxisomes, but alteration of this region interferes with nuclear localisation.

| Preparation of constructs
To create the transgenic lines CAT2 PSI , CAT2 WSQV and CAT2 ARL under the control of the CAT2 promoter, 2.4 kb genomic DNA upstream of the CAT2 gene was amplified with primers Cat2promR and Cat2prom F (all primers are listed in Table S1). The Cat2promR contains restriction sites BsrGI and XbaI. The sequence of the potential promoter region also contains HindIII at the 5′ end. The PCR product was cut with XbaI and HindIII and cloned into pGreen0179 +NosT to generate cat2promnos. CAT2 variants corresponding to CAT2 PSI CAT2 WSQV and CAT2 ARL were amplified from Arabidopsis leaf and seedling cDNA using the primer CAT2F in combination with CAT2RN, CAT2RWSQV and CAT2RARL. Correct sized amplification products were cloned into pDONOR207 by BP reaction and verified by restriction digest and sequencing. CAT2 variants were excised as BamHI and BsrGI fragments and cloned into BamHI and BsrGI digested cat2prom-nos. Finally, the native 3′UTR was placed in front of the Nos terminator. Primers PSI-3UTR-F and PSI-3UTR-R and WSQV-3UTR-F and ARL-3UTR-R were used to amplify the sequence immediately 3′ to the stop codon of the normal and splice variant transcript and to introduce flanking BsrGI and XbaI sites. PCR products and cat2promnos plasmids containing the CAT2 variants were digested with BsrGI and XbaI and ligated together to produce plasmids cat2promoter-CAT2-PSI-3UTR-nos, cat2promoter-CAT2-WSQV-3UTR-nos and cat2promoter-CAT2-ARL-3UTR-nos ( Figure S1). These plasmids were transformed into the cat2-1 mutant background by floral dip (Clough & Bent, 1998). Transformants were selected on hygromycin and three independent homozygous lines of CAT2 WSQV and CAT2 PSI and two independent homozygous lines of CAT2 ARL were used in this study. To prepare C terminal SfGFP constructs, plasmids cat2promoter-CAT2-PSI-3UTR-nos, cat2promoter-CAT2-WSQV-3UTR-nos and cat2promoter-CAT2-ARL-3UTR-nos were used as a PCR template with CAT2PSIF, CAT2PSIR, CAT2WSQVR and CAT2ARLR primers. These PCR products contain KpnI and SpeI restriction sites. After gel purification and digestion with Kpn I and Spel they were ligated into KpnI and SpeI digested PEP109E plasmid (Park et al., 2017). To prepare N terminal sfGFP plasmids cat2promoter-CAT2-PSI-3UTR-nos, cat2promoter-CAT2-WSQV-3UTR-nos and cat2promoter-CAT2-ARL-3UTR-nos were digested with BamHI and KpnI and ligated with a synthetic fragment comprising the UBQ10 promoter sequence from PEP109E, The GFP11 coding sequence and the start of the CAT2 gene from BamHI and KpnI digested pUBQ10GFP11Cat-pUC57 ( Figure S2).

| Glutathione and ascorbate quantifications
Glutathione and ascorbate were assayed as described (Noctor et al., 2016). Briefly, 100 mg of frozen leaves was ground in 1 ml perchloric acid (HClO 4 ). The homogenate was centrifuged at 14 000 rpm at 4°C and the supernatant was adjusted to pH 6.0.
Glutathione and ascorbate were quantified using Greiner F-bottom 96-well UV-transparent plates on a FLUOstar Omega microplate reader (BMG Labtech).

| Enzyme activity measurements
Approximately 100 mg leaf material was ground in liquid nitrogen.
The homogenates were centrifuged for 10 min at 14 000 rpm at 4°C, and the supernatants were used for the assay. Catalase activity assay was performed according to Veljovic-Jovanovic et al. (2001) Analysis of catalase using in-gel activity assay was carried out by using the protocol described in Weydert and Cullen (2010). Following cell disruption, three fractions were prepared as follows: The first fraction, the homogenate fraction (S1 supernatant), was obtained by centrifugation of the homogenate (10 min 2000 g). The second fraction, the cytosolic fraction (S2), was obtained by centrifugation of 1 ml of the S1 supernatant (20 min 12 000 g). Finally, the S2 pellet (organelle fraction) was suspended in 0.1 ml of the homogenisation buffer. For pure peroxisome isolation, 3-5 weeks-old wild type, cat2-1 mutant, CAT2 PSI , CAT2 and CAT2 ARL leaves (5-10 g) were harvested and leaf peroxisomes were isolated as described previously (Reumann et al., 2007). The isolation procedure was performed in a cold-room (4°C).

| Confocal microscopy analysis
Protoplasts were observed with a Zeiss 700 laser scanning confocal microscopy using EC Plan-Neofluar 20x/0.5 M27. Excitation wavelengths and emission filters were 8.5% of 488-nm diode laser/bandpass 300-531 nm to detect sfGFP. Negative controls were carried out to check that there was no bleed through of fluorescence from the chlorophyll channel.

| C-terminal modifications of AtCAT2 complement the cat2-1 mutant for growth
Previous studies have yielded contradictory results regarding the precise molecular determinants of catalase targeting, particularly the role of the carboxy-terminal amino acids. We therefore addressed this question in a physiologically relevant context where catalase activity and targeting were evaluated using the following constructs: the wild type CAT2 sequence (terminating PSI), a C terminal truncation arising from the retention of an intron that replaces the last 19 amino acids with a valine (terminating WSQV) and a variant in which the C terminal tripeptide was mutated to PTS1 consensus sequence (terminating ARL). Intron retention is the most common form of alternative splicing event (R. Zhang et al., 2017). This variant is detectable by PCR on polysomal RNA ( Figure S3) so is presumably translated but is present at very low levels in published transcriptome data sets (Calixto et al., 2018;R. Zhang et al., 2017). These constructs were expressed under the control of the native CAT2 promoter with the native 3′UTR in the cat2-1 mutant background (Queval et al., 2007) (Figure 1a). In these experiments, the transgenic lines (three independent lines for PSI and WSQV and two independent lines for ARL) were grown together with the wild type and the cat2-1 mutant for up to 5 weeks under short-day growth conditions before transfer to long-day growth conditions for a further 1 week ( Figure S4 for images of all independent transgenic lines). The cat2-1 mutant rosettes were visibly smaller than the wild type under this growth regime, as previously reported (Queval et al., 2007). However, all the lines expressing all the variants had comparable growth and development to the wild type ( Figure 1b) and rosette fresh weight was restored (Figure 1c). The leaf number and rosette diameter confirmed that the expression of all variants complemented the wild type growth phenotype ( Figures S5 and S6). The primary root length was decreased in the cat2-1 mutant, but root length was fully restored to wild type levels in all of the transgenic lines ( Figure S7).
CAT2 is a key enzyme of the photorespiratory pathway in leaves.
Flux through the photorespiratory pathway was increased by grow- 3.2 | AtCAT2 with C-terminal modifications are enzymatically active and assemble correctly The ability of all of the expressed CAT2 variants to restore the growth phenotype of the cat2-1 mutants suggests that expression of all the sequences leads to the production a functional enzyme. Leaf CAT activity was measured in three independent transgenic lines of CAT2 PSI and CAT2 WSQV and two independent transgenic lines of CAT2 ARL that had been grown under both short-and long-day conditions. CAT activities are shown in (Table S2). To enable comparisons of data from independent experiments (grey, black and white bars in 3.3 | CAT2 C-terminal variants restore cellular redox status CAT2 deficiency results in H 2 O 2 accumulation that perturbs cellular redox status, resulting in changes in the levels and oxidation state of nonenzymatic antioxidants, such as ascorbate and glutathione (Noctor & Foyer, 1998). The cat2-1 mutants had lower levels of ascorbate when grown under short-day conditions, together with an increase in the proportion of oxidised ascorbate compared to the wild type plants (Figure 3a, SD). This finding is consistent with previous reports (Mhamdi et al., 2010;Queval et al., 2007). The level of ascorbate and increased ratios of reduced to oxidised as-

| CAT2 variants target to peroxisomes
All the CAT2 variants had activity and could completely or partially (in the case of ARL) complement the various cat2-1 mutant phenotypes.
However, a key question is whether CAT2 was localised within peroxisomes. Four-week-old rosettes of wild type, cat2-1 and one of each of the transgenic lines were fractionated into a clarified homogenate (S1) a cytosol fraction (S2) and an organelle pellet (P2).
Protein and catalase activity were measured in each fraction along with hydroxypyruvate reductase (HPR) as a peroxisomal marker and expressed as activity/g fresh weight (Table 1). Peroxisomes are fragile and only 13%-42% of HPR activity was recovered in the organelle pellet. Between 41% and 78% of catalase activity was in the organelle pellet and that remaining soluble was most likely lost from organelles broken during the fractionation procedure. Calculation of the specific activity of catalase in the organelle pellet showed this was restored to wild type levels in CAT2 PSI and CAT2 WQSV but was lower in CAT2 ARL ( Figure S8). Interestingly the activity of HPR was much lower in the cat2-1 mutant compared to the wild type or complemented lines (Table 1). The reason for this is not known but catalase has been reported to have a protective effect on peroxisomal isocitrate lyase (Yanik & Donaldson, 2005) As the P2 fraction will contain other organelles, fractions of highly purified leaf peroxisomes were isolated using a combination of Percoll and sucrose density gradients (Reumann et al., 2007). In contrast to chloroplasts that were retained close to the top of the Percoll gradient, peroxisomes passed through the Percoll layer and were recovered at the bottom, as determined by analysis of the activity of the peroxisomal marker enzyme hydroxypyruvate reductase (HPR; data not shown). The isolated peroxisomes were then applied to a discontinuous sucrose density gradient. Chlorophyll was not detected in any of the fractions from the sucrose density gradient.
The distribution of CAT activity was the same as HPR, providing strong evidence for the peroxisomal location of CAT ( Figure 4). While the CAT activity of peroxisomes was very low in the cat2-1 mutants, the peroxisomes from the CAT2 PSI and CAT2 WSQV lines had similar activities to the wild type. In contrast, CAT activity was lower in the CAT2 ARL line peroxisomes, consistent with the data presented in F I G U R E 2 Restoration of catalase activity in transgenic lines and variants form active homo-tetramers. (a) Catalase activity was measured in leaf extracts from wild type, cat2-1 mutant and different independent transgenic lines grown under short (8hL 16hD) and longday (16hL 8hD) conditions and sampled at the same time relative to start of the light period. For each line, three biological replicates were each measured in triplicate. Grey, black and white bars are data from separate experiments. Percentage of catalase activity relative to wild type in each experiment is indicated at the top of the graph (activities are given in 3.5 | Dynamic localisation of catalase using self-assembling split GFP We investigated the subcellular localisation of AtCAT2 further using the self-assembling split sfGFP OPT system (Park et al., 2017), together with confocal laser scanning microscopy. For this analysis, CAT2 variants fused to sfGFP11 (at the C-terminal end) were expressed in Arabidopsis protoplasts expressing sfGFP1-10 β-strand (sfGFP1-10 OPT ) targeted to peroxisomes or nucleus ( Figure 5). Positive controls (Park et al., 2017) were used where the appropriate organelle-targeted F I G U R E 3 Redox balance and oxidative signalling are restored in complemented lines. Wild type, cat2-1 mutant, and transgenic lines were grown under short (SD; 8hL 16hD) for 4 weeks and then transferred into long-day (LD; 16h L8hD) conditions for 1 week and measurements made of reduced and oxidised ascorbate (a) and glutathione (b). In each case, black bars represent the reduced form and red bars the oxidised form. ***p < 0.001 by Student's t-test compared with the wild type (Content of the reduced form). GSSG and GSH ratio was also calculated (SD; 10:1, 2:1, 12:1 mCherry-sfGFP11 was expressed in protoplasts and labelled peroxisomes ( Figure 5a) and nucleus (Figure 5f), respectively. No GFP signal was detected in the negative control (minus a GFP11construct) and no overlap with chlorophyll florescence was observed (Figure 5b,g).
Expression of CAT2 PSI -sfGFP11 and CAT2 WSQV -sfGFP11 in protoplasts isolated from peroxisome-targeted sfGFP1-10 OPT (PX-sfGFP1-10 OPT ) reconstituted sfGFP fluorescence signal in peroxisomes (Figure 5c,d). This finding suggests that CAT is targeted to peroxisomes in these lines. In contrast, no signal was detected with CAT2 ARL (Figure 5e) with the exception of just two protoplasts in one experiment where weak fluorescence was detected in unknown membranous structures ( Figure S9). Protoplasts were also isolated from the transgenic lines expressing sfGFP1-10 OPT targeted to the nuclei (Nu-sfGFP1-10 OPT ). A reconstituted sfGFP signal was observed in the nuclei of the protoplasts transfected with CAT2 PSI -sfGFP11 construct Figure 5h). In contrast, no signal was detectable in the protoplasts transfected with CAT2 WSQV or ARL -sfGFP11 variants (Figure 5i,j).
These results were highly reproducible and were observed in multiple protoplasts in independent experiments (Table S3).
When the GFP11 was placed at the N terminus of the constructs; sfGFP11-CAT2 PSI , sfGFP11-CAT2 WSQV and sfGFP11-CAT2 ARL , a different result was obtained ( Figure 6). Peroxisomal targeting was observed when all three constructs were expressed in protoplasts derived from PX-sfGFP1-10 OPT lines (Figure 6c-e and h-j magnified image). Nuclear targeting was observed for all three variants when the N terminal tagged constructs were expressed in protoplasts derived from Nu-sfGFP1-10 OPT plants ( Figure 6m-o, bottom panel). It was previously shown (Park et al., 2017) that mCherry sfGFP11 targeted to the nucleus or peroxisome could reconstitute fluorescence in the appropriate organelle when co-expressed with cytosolic sfGFP1-10 OPT , presumably because the sfGFP11 and sfGFP1-10 can assemble in the cytosol and be subsequently imported since these organelles can import folded proteins. To determine whether the reconstitution of fluorescence in the nucleus by the sfGFP11CAT2 proteins could be due to assembly with Nu-sfGFP1-10 OPT in the cytosol followed by nuclear import, both C and N terminal tagged CAT constructs were co-expressed with CYTO-sfGFP1-10 OPT-( Figure S10). Surprisingly none of these constructs could reconstitute fluorescence in any compartment, suggesting that unlike cytosolic mCherryGFP11 the CAT constructs could not assemble with sfGFP1-10 OPT in the cytosol and that the CAT constructs are imported into the nucleus where they assemble with Nu-sfGFP1-10 OPT .

| DISCUSSION
There is little consensus regarding the importance of sequences within the conserved C terminal region of catalase in targeting to peroxisomes (Fujikawa et al., 2019;Kamigaki et al., 2003;Mullen et al., 1997). Previous studies have used different approaches that rendered interpretation of the findings difficult. To resolve this problem, we addressed the question using a system that is as physiologically relevant as possible, by expressing untagged AtCAT2 variants under the native CAT2 promoter in the cat2-1 mutant. This allowed testing not only of catalase targeting but crucially of functionality. Our data clearly demonstrate that the last 18 amino acids of CAT2 are not required for activity, peroxisome location or ability to complement any of the cat2-1 mutant phenotypes under any of the growth conditions tested.
Intriguingly the mutation of the terminal 3 amino acids of wild type CAT2 -PSI-COOH to a PTS1 consensus sequence -ARL-COOH led to only a partial complementation of some phenotypes. Whilst the CAT2 ARL variant was assembled into tetramers and targeted to peroxisomes, its activity was reduced. This is consistent with the finding that exchanging the C-terminal tripeptide SKI of Hansenula polymorpha catalase for the consensus PTS1 sequence SKL resulted in reduced catalase activity in cell lysates and the formation of catalase aggregates in peroxisomes, leading the authors to propose that a lower affinity of the catalase PTS1 for PEX5 resulted in slower import, allowing time for catalase maturation in the cytosol (Williams et al., 2012). The folding and maturation pathway of catalase remains unclear. To produce a mature catalase protein, the subunits have to fold, bind haem and tetramerise. In A. thaliana, the NCA1 protein interacts with catalase in the cytosol to promote the formation of active enzyme; nca1 mutants lack catalase activity and are hypersensitive to multiple stresses  and immunity mediated autophagy (Hackenberg et al., 2013).
The H. polymorpha catalase C terminal sequence SKI binds PEX5 with 8-fold reduced affinity compared to an equivalent peptide with the C terminal sequence SKL (Williams et al., 2012). In contrast, a peptide corresponding to the C terminus of AtCAT2 showed no F I G U R E 5 C terminal tagged wild type CAT2 and short-form target to peroxisomes, wild type can also target to nucleus. CAT2 variants tagged with sfGFP11 at the C terminus transfected into protoplasts of transgenic Arabidopsis expressing sfGFP1-10 OPT targeted to peroxisome (top) and nucleus (bottom  (Oshima et al., 2008).
S. cerevisae peroxisomal catalase Cta1p is also independent of the PEX5 TPR domain for import (Rymer et al., 2018), and mammalian catalase interacts with the N terminal part of PEX5 and this interaction blocks catalase tetramerisation (Freitas et al., 2011 The above discussion leads to the question of why the C-terminal 18 amino acids of CAT2 is so highly conserved, if the sequence is apparently dispensable. Cell fractionation is not appropriate for studying dynamic changes in protein localisation so the superfolder split GFP system was employed. This system was developed to study translocation of type III secretion system substrates which are folded and whose translocation is often prevented when fused to large fluorescent proteins (Park et al., 2017). The 11th β strand of GFP (GFP11) is only 16 amino acids, not much larger than an epitope tag, and should be minimally invasive when fused to a target protein.
F I G U R E 6 N terminal tagged CAT2 variants target to nucleus and peroxisomes. CAT2 variants tagged with sfGFP11 at the N terminus transfected into protoplasts of transgenic Arabidopsis expressing sfGFP1-10 OPT targeted to peroxisome (a-j) and nucleus (k-o). The middle row (f-j) shows a magnified image of the protoplast indicated by a white dotted box in the panel above. Positive control: PX-mCherry-11 (a,f) and NU-mCherry-11 (k) for peroxisome and nucleus, respectively. When GFP11 was fused to the C terminus of either CAT2 PSI or CAT2 WSQV and expressed in protoplasts derived from plants with GFP1-10 targeted to peroxisomes, these constructs targeted to peroxisomes demonstrating that a free C terminus was not required.
Conversely, CAT2 ARL was not targeted to peroxisomes when GFP11 was appended to the C terminus, as might be expected if the protein now depends upon canonical PTS1-PEX5 interaction which requires a free carboxylate on the C terminal leucine residue. However, it appears not to be able to access the pathway used by CAT2 PSI and CAT2 WSQV . Finally, the 3 catalase variants were co-expressed with nuclear-targeted GFP1-10. CAT2 PSI was efficiently targeted to the nucleus whether the sfGFP11 was at the N or C terminus but CAT2 ARL and CAT2 WQSV could only target to the nucleus when the sfGFP11 was at the N terminus, suggesting that alteration of the C terminus in combination with a C terminal tag interferes with the recognition by some unknown factor that promotes nuclear targeting ( Figures 5 and 6).
Catalases can be relocated to the nucleus through the interaction with effector proteins of Phytopthora sojae to manipulate PCD (M. Zhang et al., 2015) and the cucumber mosaic virus (CMV) protein 2b leading to viral-induced necrosis (Inaba et al., 2011;Murota et al., 2017). An effector protein from the plant growth promoting rhizobacterium Saccharothryx yanglingensis interacts with catalase in the nucleus and stimulates plant immunity (Y. Zhang et al., 2018).
Thus, redirection of catalase to the nucleus appears to be a strategy employed by microorganisms to modulate the outcomes of their interactions with plants. Our results demonstrate that catalase has the capacity to be transported to the nucleus independent of any exogenous effector proteins. This is a specific response and not a consequence of cytosolic assembly with nuclear targeted GFP1 −10 followed by import since CAT2-WSQVsfGFP11 and CAT2-ARLsfGFP11 were not nuclear. Furthermore, CAT2 constructs were not able to assemble with cytosolic GFP1-10 which argues for their autonomous import and subsequent assembly with nuclear-targeted GFP1-10. This finding may explain the reported interaction of catalase with nucleoredoxin 1 (NRX1; Kneeshaw et al., 2017) and suggests that catalase dynamics could be a component of innate plant responses and that the highly conserved C terminal region plays an important role in this behaviour.