Regulation of heat shock proteins 70 and their role in plant immunity

The extensive regulation of HSP70 has received little attention in plant science. Here, we summarize current knowledge gained in large-scale profiling experiments and highlight HSP70s’ role in plant immunity.


Introduction
Heat shock proteins (HSPs) were discovered by Italian scientist Ferruccio Ritossa who noticed a rapid transcriptional response to heat shock in Drosophila melanogaster and eventually published his results in 1962 (see Ritossa, 1996). The term 'heat shock proteins' was adopted 12 years later (Tissiéres et al., 1974). What are HSPs? These ubiquitous and widespread proteins across prokaryotic and eukaryotic organisms were first discovered in response to an increase in temperature, but accumulated evidence indicates that these proteins are involved in much more than that. Foremost, HSPs act as chaperones and are involved in the assembly, stabilization, and maturation of proteins and protein complexes (Wang et al., 2004). Besides their chaperone functions, HSPs play a role in plant development and response to abiotic and biotic stress conditions, including drought, salinity, pathogen infection, and insect attacks (Park and Seo, 2015;Usman et al., 2017;Ul Haq et al., 2019). The HSP family covers a variety of proteins, ranging from 10 kDa to >100 kDa, and the molecular weight is the basis of present-day HSP nomenclature (Ratheesh Kumar et al., 2012). Several recent publications have reviewed the role of plant HSP90 (Tichá et al., 2020), HSP40 , and small HSPs (Asea et al., 2016;Waters and Vierling, 2020). Here, we will outline the role of HSP70s, one of the most studied HSPs in animals, and discuss their somewhat overlooked roles in plants.

HSP70 structure is evolutionarily conserved
HSP70s are highly conserved ATP-dependent chaperones, and this family shows the highest sequence similarity within a large evolutionary distance (Borchiellini et al., 1998). An example is DnaK protein (prokaryotic HSP70), which shares ~50% amino acid identity with eukaryotic HSP70s. The genetic conservation of the HSP70 sequence is also evident in Arabidopsis accessions. The frequency of mutations is low, seems to correlate with geographical distribution (Fig. 1A), and 10 proteins show zero high-impact mutations (Table 1). Plant HSP70s share at least two of the four structural features of the archetype bacterial HSP70 DnaK: an N-terminal, 45 kDa nucleotide-binding domain (NBD), followed by a 15 kDa substrate-binding domain (SBDβ), a 10 kDa helical lid domain (SBDα), and a disordered C-terminal tail of variable length ( Fig. 1C; Sarkar et al., 2013). In eukaryotic cytosolic and nuclear HSP70s, the disordered  (Weigel and Mott, 2009) and visualized using a Microsoft Excel 3D map. (B) Domain organization of DnaK and 110/SSE subfamilies (Lin et al., 2001;Sarkar et al., 2013). (C) Structure of HSP70 with bound ATP. Modeled using the crystal structure of bacterial DnaK (PDB ID: 4B9Q;Kityk et al., 2012). Red, nucleotide-binding domain (NBD); blue, substrate-binding domain SBDα; violet, substrate-binding domain SBDβ; gray, ATP; yellow, linker. Sites of protein-protein interactions with HSP70 co-chaperones are indicated, including tetratricopeptide repeat (TPR) proteins, J-domain proteins (JDP), and nucleotide exchange factors (NEF). (D) Chaperone cycle of HSP70s. The interaction with JDPs stimulates ATP hydrolysis and the transition to the ADP-bound state with a high affinity for protein substrates. NEFs facilitate the HSP70 conversion back to the low-affinity state, the dissociation of ADP and rebinding of ATP, and the release of the protein substrate (Rosenzweig et al., 2019). Based on UniProt (Bateman et al., 2017), SUBA 4.0 (Hooper et al., 2017), the 1001 genome database (Weigel and Mott, 2009) C-terminal tail frequently ends with a characteristic charged motif (EEVD) that interacts with specific co-chaperones containing TPR repeats, including HSP-organizing protein (HOP; transfer of unfolded HSP70 substrates to foldases; Rosenzweig et al., 2019), HSP-interacting protein (HIP; stabilization of the high-affinity ADP-bound state of HSP70; Zuiderweg et al., 2017), and C-terminus of HSP70-interacting protein (CHIP; ubiquitination mediating degradation of unfolded HSP70 substrates via the proteasome; Fernández-Fernández et al., 2017).

Chaperone activity of HSP70s
The mediation of protein folding and translocation is the bestdescribed function of HSP70s. The promiscuity of the HSP70 interaction and efficient targeting to substrate proteins are based on a transient interaction of the SBD domain ( Fig. 1B-D) with a short degenerative sequence motif that is frequently found in virtually all proteins (Rüdiger et al., 1997;Kityk et al., 2012). The HSP70-substrate interaction depends on a complicated allosteric mechanism that couples ATP hydrolysis in the NBD with substrate capture by the SBD (described in more detail in Qi et al., 2013;Kityk et al., 2015;Rosenzweig et al., 2019). The HSP70 binds to protein substrates with a significantly increased affinity in the ADP-bound state, and ATP hydrolysis is essential for its chaperone function (Kityk et al., 2018). However, the intrinsic ATPase rate is very low (Kityk et al., 2012) and requires co-chaperones, namely J-domain proteins (JDPs; accelerate ATP hydrolysis) and nucleotide exchange factors (NEFs; accelerate ADP-ATP exchange) (Abrams et al., 2014;Bracher and Verghese, 2015;Zuiderweg et al., 2017). The whole chaperone cycle is illustrated in Fig. 1D. In contrast to other chaperones, HSP70s do not require self-interaction for the chaperone functions (Takakuwa et al., 2019). Besides the described interactions and chaperone complexes that facilitate protein folding, HSP70s are essential for driving the movement of proteins across the membrane and protein translocation into the endoplasmic reticulum (ER), mitochondria, and plastids (Su et al., 2010;Schleiff and Becker, 2011;Nakai, 2018;Sun et al., 2021). Translocases have narrow channels, limiting the transport to a completely unfolded chain or, at most, α-helices (Craig, 2018). Interestingly, organellar HSP70s are more critical for the translocation process than their cytosolic counterparts, and form the core of the so-called import motors. There are different models for the HSP70 motor action mechanism that pulls the protein into an organelle (e.g. Sousa and Lafer, 2019), but the substrate-HSP70 interaction itself is identical to that of the standard chaperone cycle and requires JDP-driven ATP hydrolysis.

HSP70s mediate quality control
HSP70s have an essential role in protein quality control. The fate of misfolded and aggregated proteins is decided by a combination of HSP70 and its interactors, and the defective protein is targeted for degradation through the ubiquitin-proteasome system, as well as through autophagy pathways. The most frequent means of protein degradation in eukaryotic cells requires the ubiquitin-26S proteasome system. Proteins are targeted via ubiquitination through the sequential action of three discrete enzymes: the ubiquitin-activating enzyme (E1), ubiquitinconjugating enzyme (E2), and ubiquitin ligase (E3). The E3 ligase is responsible for substrate recognition, and a highly conserved and chaperone-dependent E3 ligase is CHIP. This enzyme interacts with HSP70 and ubiquitinates protein substrates captured by HSP70. CHIP has been extensively studied in many model organisms, including plants (reviewed in, for example, Zhang et al., 2021), and accumulated knowledge suggests that it is an integral component of both biotic and abiotic stress control. Chaperone-mediated autophagy is mediated by cytosolic HSP70s and targets proteins containing the signal motif KFERQ (Fernández-Fernández et al., 2017). This was the first described chaperone-assisted pathway but requires a receptor that seems to be missing in plant lineages (Lescat et al., 2020). Similarly, the autophagy pathway mediated by endoplasmic HSP70 has not yet been confirmed in plants (Cha-Molstad et al., 2015).

Classification and localization of plant HSP70s
The plant HSP70 protein superfamily consists of two subfamilies: HSP70 (DnaK) and HSP110/SSE (Fig. 1B). The Arabidopsis genome encodes 14 and four members of HSP70 and HSP110/SSE subfamilies, respectively. HSP110/SSE exhibits homology to yeast SSE1 and human HSP110. The HSP110 sequence contains an insertion of acidic residues in the substrate-binding domain, an extension of the C-terminal domain, and the resulting protein is significantly larger (Fig. 1B).
All eukaryotes possess multiple HSP70s. In plants, four distinct sets function in the cytosol, mitochondria, chloroplasts, and endoplasmic reticulum (Table 1). This localization is determined by conserved N-and C-terminal sequences and is often used to divide HSP70s into four major subgroups: cytosolic/nuclear (EEVD motif), ER (HDEL motif), plastidic (PEGDVIDADFTDSK motif), and mitochondrial (PEAEYEEAKK motif) (Guy and Li, 1998). Those targeted to the same compartment share a similar evolutionary history. HSP70s targeted to the ER and cytosol resulted from gene duplication and subsequent divergence, and HSP70s targeted to mitochondria and chloroplasts evolved by gene transfer from endosymbionts and are more related to the prokaryotic DnaK (Waters and Rioflorido, 2007;Huang et al., 2008) The low correlation between gene expression and protein abundance indicates extensive posttranscriptional or post-translational control of HSP70s The analysis of available Arabidopsis protein data and expression profiles showed that only a minor fraction of HSP70s has a positive linear correlation between transcript and protein abundance ( Fig. 2A). A higher correlation (Spearman correlation coefficient r >0.7) was found only for HSP70-4, HSP70-5, and HSP70-8. The highest deviations were found for HSP70-18 (expression data available only for imbibed seeds) and HSP70-10 (not found in the comparative dataset stored in the Arabidopsis tissue atlas). Translation elongation rates, post-transcriptional regulations, post-translational modifications, and protein life span impact the resulting protein abundances. Protein degradation rates have been estimated for seven Arabidopsis HSP70s (L.  and show significant differences. The lowest average degradation rates were found for plastid-localized isoforms HSP70-6 (0.069 d -1 ) and HSP70-7 (0.082 d -1 ), indicating an intermediate protein turnover rate. In contrast, cytosolic HSP70s showed a rapid degradation, with an average degradation rate of HSP70-14 reaching 1.878 d -1 . Information about the post-transcriptional control of plant HSP70s is missing, but it has been demonstrated that HSP70 pseudogenes, including products of HSP70 alternative splicing, may act as long non-coding RNAs and interfere with the HSP70 translation (Bernabò et al., 2020). Arabidopsis HSP70-18 was considered a pseudogene, but the proteome analyses showed that this is not the case ( Fig. 2A). There are 35 predicted HSP70 gene models, and only eight Arabidopsis HSP70 genes are missing putative alternatively spliced variants (Table 1). It is possible that at least some of these variants could participate in transcriptional control, but the sequences are highly similar. Except for AT1G79920.2, the available peptidebased evidence cannot confirm or exclude the existence of the corresponding proteins (van Wijk et al., 2021, Preprint).

Plant hormones control HSP70 expression
The phytohormones are small-molecule regulators that collectively regulate plant growth and development, and facilitate the integration of environmental cues, including biotic and abiotic stressors. It is thus not surprising to find Arabidopsis HSP70 among the list of phytohormone-responsive genes and proteins (Fig. 2B, C). Only two cytosolic HSP70s appear to be missing in response to phytohormones, namely HSP70-5 and HSP70-8. Somewhat counterintuitive is the reported response to hormones traditionally associated with a stress response. Abscisic acid, jasmonates, and ethylene showed a predominantly negative effect on HSP70 expression and the abundances of the corresponding proteins. Arabidopsis data for salicylic acid treatment were not available in the selected datasets, but previous proteomic analyses showed a decrease in HSP70 after salicylic acid treatment of Cucumis sativus and Cunninghaimia lanceolata (Černý et al., 2016). In contrast, cytokinin up-regulates most HSP70s and positively impacts HSP70 abundances (Fig. 2B). That could coincide with the predicted role of cytokinin in response to heat shock (Černý et al., 2014). Interestingly, abscisic acid and cytokinin antagonistically regulate many developmental processes (Huang et al., 2018), and the observed antagonistic effect on HSP70s (Fig. 2B, C) indicates that its regulation could be a part of the crosstalk between these two phytohormones.

HSP70s are subject to extensive post-translational control
The Plant PTM viewer lists almost 300 HSP70 peptides with a post-translational modification (Willems et al., 2019). The most frequent modifications are phosphorylation (65 peptides), acetylation (53 peptides), sulfenylation (33 peptides), and ubiquitination (27 peptides). However, these are the most intensively studied plant protein modifications, and the apparent enrichment probably does not reflect the true proportion of different modifications. As illustrated in Fig. 3, the sites of these modifications are highly conserved and most can be mapped onto the consensus of HSP70 family sequences. Experimental evidence in plant protein science is largely missing, but HSP70 modifications have been established as the so-called 'chaperone code' and can determine protein localization, selectivity, activity, and protein-protein interactions (Cloutier and Coulombe, 2013;Nitika et al., 2020). These modifications represent both the fine-tuning of HSP70 biology and a means of rapid response to stress . The activation of HSP70 expression is rapid, but corresponding protein biosynthesis is significantly slower. For instance, the delay in heat-induced accumulation of Arabidopsis HSP70-4 and HSP70-5 is >2 h compared with the corresponding transcripts (Palmblad et al., 2008). Post-translational modifications are thus the first line of defense. Phosphorylation and acetylation have been shown to regulate the oligomerization of HSP70s (Morgner et al., 2015). The dimerization of HSP70s effectively buries its binding domains and inhibits protein-protein interactions (Takakuwa et al., 2019), and thus the regulation of the monomer/oligomer pool represents a rapid control mechanism to regulate the activity of HSP70s.
The Arabidopsis HSP70 sequences contain on average 1.2% cysteines, and sulfenylation and nitrosylation have been reported for these residues (Fig. 3). Cysteine is the most reactive proteinogenic amino acid, and thiols often act as a redox molecular switch regulating protein activity (e.g. Černý et al., 2013). In fact, sulfenylation of HSP70 has been demonstrated to activate heat shock transcription and induce thermotolerance in yeast . Interestingly, HSP70 may also directly participate in reactive oxygen metabolism by regulating the transport of superoxide dismutase into mitochondria (Zemanovic et al., 2018).

Role of HSP70s in plant immunity
Analysis of publicly available data stored in the Arabidopsis RNA-Seq Database  showed that most  (Waterhouse et al., 2009) and (E) Arabidopsis HSP70 genes in response to biotic interaction and flagellin peptide fragment (FLG-22; elicits defense response). The ratio represents the median log2 expression compared with mock-treated plants. Expression data were retrieved from the Arabidopsis RNA-Seq Database .
HSP70 genes are up-regulated in response to biotic stressors (Fig. 2E). Only cytosolic HSP70-8 and chloroplastic HSP70-6 and HSP70-7 were predominantly repressed, and expression profiles for HSP70-18 were unavailable. Comparison of the expression profile and sequence feature similarity (Fig. 2D, E) did not show any significant similarity in the clusterings (adjusted Rand index 0.16), indicating that the response to biotic stress is not correlated with the HSP70 sequence. However, the contrasting response of chloroplastic HSP70 genes and the available literature indicate that HSP70 localization is the key factor.
Plants respond to pathogen invasion using a two-branched innate immune system consisting of PAMP (pathogenassociated molecular pattern)-triggered immunity (PTI) and effector-triggered immunity (ETI; Abdul Malik et al., 2020). PTI represents an early response mechanism mediated by pattern recognition receptors (PRRs) that recognize and bind PAMPs (Nürnberger and Kemmerling, 2009). ETI is a reaction to pathogen-secreted effectors interfering with PTI. This second defensive barrier recruits resistance (R) proteins that specifically recognize pathogen effectors and usually triggers a hypersensitive response and cell death (Abdul Malik et al., 2020). HSP70s are an integral part of plant immunity and participate in both PTI and ETI responses (Fig. 4).

Cytosolic HSP70s regulate ETI
The HSP70 subfamily (DnaK) members with a C-terminal EEVD motif (sometimes referred to as heat shock cognate 70 chaperones) regulate immune responses in plants by cooperating with cytosolic chaperones such as SGT1, HSP90, and RAR1. This interaction is an integral part of the processing and folding of NB-LRR (nucleotide-binding domain-leucine-rich repeat) immune sensors (Shirasu, 2009). There are four cognate HSP70 genes in Arabidopsis (isoforms 1-4), and three of these are strongly up-regulated in response to different pathogens (Fig. 2E). Limited evidence indicates that these HSP70 genes are at least partially redundant, and experiments with a loss-of-function mutation of individual HSP70 isoforms 1, 2, and 3 did not reveal any difference in the resistance against pathogens (Noël et al., 2007; Table 1). Surprisingly, HSP70-1 is mostly unresponsive in biotic stress experiments ( Fig. 2E; Table 2). A transgenic line overexpressing HSP70-1 has a reduced NB-LRR-dependent immunity response (Noël et al., 2007), but that could coincide with an attenuated expression of five different HSP70 genes (Sung and Guy, 2003). The HSP70 interactor SGT1 is a co-chaperone of HSP90 and, in addition to ETI, also regulates auxin and jasmonic acid signaling by maintaining steady-state levels of the corresponding receptors TIR1 and COI1 (X.-C. Wang et al., 2016). It has been proven that the COI1 stabilization is facilitated by SGT1b-HSP70-HSP90 chaperone complex (X.-C. , and HSP70s thus directly influence hormonal perception. The critical role of cytosolic HSP70s in plant defense is supported by the fact that these proteins are targeted by pathogen effector proteins. HopI1, a virulence effector of pathogenic Pseudomonas syringae, binds directly to the host HSP70 and recruits it to chloroplasts, the site of HopI1 localization (Jelenska et al., 2007(Jelenska et al., , 2010. The interaction was confirmed by co-immunoprecipitation for all four Arabidopsis isoforms (1-4) and for chloroplastic HSP70-6. Further, plants with depleted cytosolic HSP70-1 were more susceptible to P. syringae (Jelenska et al., 2010), and silencing of cytosolic HSP70 in pepper (Capsicum annuum) increased the plant's susceptibility to Xanthomonas campestris (Kim and Hwang, 2015). Finally, the effector Pi23226 of Phytophthora infestans was co-immunoprecipitated with HSP70s, and transient overexpression of these HSP70s in Nicotiana benthamiana inhibited Phytophthora growth (Lee et al., 2018).  (Waterhouse et al., 2009), visualized by Protter (Omasits et al., 2014), and projected into the 3D dimeric structure by SWISS-MODEL (Waterhouse et al., 2018). Data were retrieved from the Plant PTM Viewer database (Willems et al., 2019).

HSP70s resident in the endoplasmic reticulum are related to endoplasmic reticulum-mediated plant immunity
Recent studies have pointed out the ability of plant pathogens to utilize their effectors to bind to the host ER stress pathway and manipulate it to their advantage (reviewed, for example, in Jing and Wang, 2020). ER stress occurs when the ER's ability to fold proteins becomes saturated and can induce programmed cell death (PCD; Eichmann and Schäfer, 2012). Cells have evolved an unfolded protein response mechanism that induces the expression of genes that encode ER chaperones, including HSP70 genes. ER-resident HSP70s (also called binding protein, BiPs) participate in plant immunity as a central regulator and are the link to ER-supported immunity functions. BiP genes are induced by salicylic acid (Nagashima et al., 2014;Poór et al., 2019) and multiple biotic stressors (Figs 2E,4B), and the expression is reportedly regulated by NPR1, the master regulator of systemic acquired resistance (Wang et al., 2005). BiPs stabilize folding intermediates, prevent aggregation, and aid in the subsequent protein folding and assembly (Padmanabhan and Dinesh-Kumar, 2010). In contrast to mammals and yeast, flowering plants contain several BiP proteins (HSP11-13 in Arabidopsis; Noh et al., 2003). BiP1 and BiP2 are highly similar based on their amino acid sequence, and they are ubiquitously expressed under non-stressed conditions, whereas BiP3 is more distantly related, highly induced upon ER stress, and therefore classified as a marker gene of the Arabidopsis unfolded protein response (Strasser, 2018). Several studies have shown that a process called ER quality control is essential for the proper function of the pattern recognition receptors, and part of this pathway relies on the BiP chaperone complex (Anelli and Sitia, 2008;Strasser, 2018). Nekrasov et al. (2009) reported that a subset of LRR-receptor-like kinases (RLKs), including EF-Tu (elongation factor thermounstable) receptor (EFR), require one or several ER complexes comprising SDF2, ERdj3B (Hsp40 co-chaperone), and BiP for their accumulation and subcellular localization. This chaperone complex regulates the biogenesis of multiple components of the plant defense system, including Cf proteins (resistance to the foliar pathogen Cladosporium fulvum; Liebrand et al., 2012) and Ve1 (resistance to the vascular fungal pathogen Verticillium dahliae; Liebrand et al., 2014). In contrast to the positive effect of cytosolic HSP70 overexpression, BiP3 overexpression compromised immunity against Xanthomonas oryzae (Park et al., 2010(Park et al., , 2014. OsBiP3 can bind to the critical XA21 receptor that facilitates resistance, and the presence of high levels of BiP reduces its stability and function (Park et al., 2010). The silencing of the individual BiPs did not affect EFR or Cf function, but a knockdown of the different BiPs in tomatoes resulted in a compromised Ve1-mediated resistance to V. dahliae (Nekrasov et al., 2009;Liebrand et al., 2012Liebrand et al., , 2014. The bip2 mutant showed compromised secretion of PR1, leading to an impaired resistance against the bacterial pathogen Pseudomonas syringae (Wang et al., 2005).
HSP70s were found to play a role in developmentally regulated PCD (Rowarth et al., 2020), and several studies suggest that BiPs are responsible for that (see, for example, Zhang et al., 2014). Furthermore, this mechanism is utilized by both mutualistic fungi and pathogens. For example, the Arabidopsis mutant bip2 showed enhanced fungal colonization rates compared with the wild type, and BiP protein levels were reduced in colonized wild-type roots compared with non-colonized roots, suggesting that impaired BiP accumulation supports fungal development during cell death-associated colonization (Qiang et al., 2012). Similarly, silencing of BiP resulted in a delay of the hypersensitive response and PCD induced by Xanthomonas oryzae pv. oryzae (Xu et al., 2012), and BiP-overexpressing lines displayed an accelerated hypersensitive response triggered by Pseudomonas syringae pv. tomato in soybean and tobacco (Carvalho et al., 2014). It has also been found that the effector PsAvh262 secreted by Phytophthora sojae stabilizes BiPs and suppresses ER stress-triggered cell death (Jing et al., 2016). Viruses can also exploit this pathway (Gaguancela et al., 2016), and up-regulation of BiP was found in many previous studies, including response to Potato virus X, Turnip crinkle virus, Garlic virus X, and Sugarcane mosaic virus (Ye et al., 2011;Lu et al., 2016;Moon et al., 2016;Chen et al., 2017). A recent study showed that the BiP accumulation inhibits the ER to nucleus translocation of Bcl-2associated athanogene (BAG7; Zhou et al., 2021). This protein is a member of a multifunctional group of co-chaperones, and its role in heat tolerance, unfolded protein response, and basal immunity is well established (Williams et al., 2010;Li et al., 2016;. The accumulated evidence indicates that the BAG7 translocation to the nucleus is critical for its protective function and that its BiP-induced retention in the ER promotes susceptibility to pathogens (Zhou et al., 2021).

HSP70 accumulation and plant resistance to pathogens
The reported changes in HSP70 expression and protein accumulation show a predominantly positive response to biotic stimuli ( Fig. 5; Table 2), and many reports have indicated that an increase in the plant pathogen resistance positively correlates with the HSP70 accumulation. For instance, the comparison of tomato accessions showed that the accession resistant to Phytophthora parasitica had a significantly higher HSP70 expression (Naveed et al., 2018). Similarly, faster induction of HSP70 genes was correlated with enhanced resistance to Blumeria graminis in barley (Molitor et al., 2011). In contrast, at least some pathogens exploit the host's HSP70 machinery, and a loss-offunction mutation or silencing of HSP70 reportedly stimulate the resistance of the plant to viruses (reviewed in, for example, Hýsková et al., 2021a). However, the effect of mutation is not that simple in a partially redundant HSP70 gene family, and the interpretation should be carefully evaluated. The analysis of gene expression of Arabidopsis hsp70-15 knockout plants showed that the mutation significantly increased HSP70-4 (14-fold change) and HSP70-2 (6-fold change) (Jungkunz et al., 2011). These members of the DnaK subfamily are the most frequently found HSP70s in the biotic response, and the observed mutant resistance to the Turnip mosaic virus could coincide with that.

Conclusion and future perspectives
Plant HSP70 research is lagging behind its animal counterparts, and our knowledge about post-transcriptional and posttranslational control is only rudimentary. It is obvious that this should be the next objective in plant HSP70 research. Animal HSP70s play a significant signaling role in the extracellular space, particularly in the inflammatory and immune responses (see, for example, Mambula et al., 2007). According to available data and similar to its animal counterparts, none of the Arabidopsis HSP70s contains a targeting signal for secretion. Despite the presence of cell walls, plants can secrete proteins without the consensus N-terminal peptide sequence via  Table 2. Localization is based on the expected localization of Arabidopsis HSP70s and corresponding putative Arabidopsis orthologs. The color of the cellular compartments represents the predominant response of the resident HSP70s to biotic stimuli. Red, blue, and purple represent a positive, negative, and mixed response to biotic stimuli, respectively. alternative, unconventional protein secretory pathways into the extracellular space. The exact extent of these alternative pathways in plants has not yet been fully elucidated, and the role of plant HSP70s in the extracellular space (if any) is unknown. However, at least nine Arabidopsis HSP70s have been detected in the extracellular space (Table 1), and it has been confirmed that the extracellular vesicles in the plant secretome contribute to the plant defense system (Regente et al., 2017). The role of extracellular HSP70 could be evolutionarily conserved and could provide an additional explanation for the observed accumulation in response to pathogens. Finally, our review highlighted a prominent role for the plant hormone cytokinin in the regulation of HSP70 genes. Cytokinin promotes plant resistance against pathogens such as bacteria, fungi, and pest insects (Akhtar et al., 2020), and the up-regulation of HSP70 genes is probably an overlooked part of that effect.

Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding
Funding support for this work was provided by the Ministry of Education, Youth and Sports of the Czech Republic (grant no. CZ.02.1.0 1/0.0/0.0/16_019/0000738) with support from the European Regional Development Fund -Project 'Centre for Experimental Plant Biology' (MC, RK, and BB), and by the Ministry of Education, Youth and Sports of the Czech Republic (grant no. CZ.02.2.69/0.0/0.0/19_073/001667 0) with support from the European Social Fund -Project 'Internal grants of Mendel University in Brno' (MB, RK, VB, and MC).
Chen X, Shi L, Chen Y, Zhu L, Zhang D, Xiao S, Aharoni A, Shi J, Xu