Overexpression of a soybean nuclear localized type–III DnaJ domain-containing HSP40 reveals its roles in cell death and disease resistance


  • Jian-Zhong Liu,

    1. Department of Plant Pathology and Microbiology, Iowa State University, Ames, Iowa, USA
    2. College of Chemistry and Life Sciences, Zhejiang Normal University, Jinhua, China
    Search for more papers by this author
  • Steven A. Whitham

    Corresponding author
    • Department of Plant Pathology and Microbiology, Iowa State University, Ames, Iowa, USA
    Search for more papers by this author

For correspondence (e-mail swhitham@iastate.edu).


Heat-shock proteins such as HSP70 and HSP90 are important molecular chaperones that play critical roles in biotic and abiotic stress responses; however, the involvement of their co-chaperones in stress biology remains largely uninvestigated. In a screen for candidate genes stimulating cell death in Glycine max (soybean), we transiently overexpressed full-length cDNAs of soybean genes that are highly induced during soybean rust infection in Nicotiana benthamiana leaves. Overexpression of a type-III DnaJ domain-containing HSP40 (GmHSP40.1), a co-chaperone of HSP70, caused hypersensitive response (HR)-like cell death. The HR-like cell death was dependent on MAPKKKα and WIPK, because silencing each of these genes suppressed the HR. Consistent with the presence of a nuclear localization signal (NLS) motif within the GmHSP40.1 coding sequence, GFP-GmHSP40.1 was exclusively present in nuclear bodies or speckles. Nuclear localization of GmHSP40.1 was necessary for its function, because deletion of the NLS or addition of a nuclear export signal abolished its HR-inducing ability. GmHSP40.1 co-localized with HcRed-SE, a protein involved in pri-miRNA processing, which has been shown to be co-localized with SR33-YFP, a protein involved in pre-mRNA splicing, suggesting a possible role for GmHSP40.1 in mRNA splicing or miRNA processing, and a link between these processes and cell death. Silencing GmHSP40.1 enhanced the susceptibility of soybean plants to Soybean mosaic virus, confirming its positive role in pathogen defense. Together, the results demonstrate a critical role of a nuclear-localized DnaJ domain-containing GmHSP40.1 in cell death and disease resistance in soybean.


DnaJ proteins, also known as HSP40s (heat-shock protein 40), are a family of conserved co-chaperones for HSP70s (Caplan et al., 1993; Silver and Way, 1993; Qiu et al., 2006). DnaJ proteins contain a 70-amino-acid consensus sequence known as the J domain, through which they interact with HSP70 proteins (Hennessy et al., 2005). HSP70s have a weak ATPase activity that is stimulated by interaction with DnaJ proteins (Fan et al., 2003). The enhanced ATPase activity of HSP70 is required for the stable binding and proper folding of its client proteins (Caplan et al., 1993). HSP70 proteins are involved in many cellular processes, including protein folding, protein translocation across membranes and regulation of protein degradation. Thus, DnaJ proteins play critical roles in a variety of cellular processes (Liberek et al., 1991; Scidmore et al., 1993; Cyr et al.,1994; Goffin and Georgopoulos, 1998). Besides their co-chaperone activity, DnaJ proteins also function as a protein disulfide isomerase to catalyze the formation, reduction or isomerization of disulfide bonds (de Crouy-Chanel et al., 1995).

The J domain, the hallmark of J proteins, comprises four α-helices and an invariable tripeptide of His, Pro and Asp (the HPD motif; Rajan and D'Silva, 2009). J proteins have been classified into three types (I, II and III) based on the presence of other conserved domains (Miernyk, 2001; Walsh et al., 2004; Rajan and D'Silva, 2009). Traditional type-I J-domain proteins contain a J domain, a Gly/Phe-rich domain (G/F), a CXXCXGXG zinc-finger domain and a less conserved C-terminal domain. Type-II J proteins lack the zinc-finger domain, whereas type-III J proteins contain only the J domain. Type-IV J proteins have been recently described and classified as ‘J-like proteins’, with significant sequence and structural similarities with the J domain, but they lack the HPD motif (Walsh et al., 2004).

A total of 120 J-domain proteins have been identified in Arabidopsis. These J-domain proteins have been reported to localize in different subcellular compartments and participate in various biological processes (Miernyk, 2001; Rajan and D'Silva, 2009). Bioinformatic analysis predicts that 50 J proteins are localized in the cytosol, 19 in mitochondria, 12 in chloroplasts, nine in the endoplasmic reticulum (ER), three in the cytoskeleton, one in the plasma membrane, 24 in the nucleus and two in the vacuole (Rajan and D'Silva, 2009). However, only a few of them have been functionally characterized (Park et al., 2011; Shen et al., 2011).

Plant J-domain proteins have been reported playing diverse roles in stress responses (Wang et al., 2004; Ham et al., 2006; Yang et al., 2010), developmental processes such as flowering time and male sterility (Tamura et al., 2007; Kneissl et al., 2009; Yang et al., 2009; Park et al., 2011; Shen et al., 2011), and in chloroplast movement (Suetsugu et al., 2005). In host–pathogen interactions, viral coat and movement proteins interact with J proteins to facilitate viral assembly and movement, respectively (Hofius et al., 2007; Shimizu et al., 2009). Silencing NtMPIP1, a tobacco DnaJ-like protein, significantly inhibited the spread of TMV (Shimizu et al., 2009). Interestingly, a recent study revealed that virulence effector HopI1, a chloroplast-targeted class-III J protein from Pseudomonas syringae, suppresses both salicylic acid (SA) accumulation and host defense responses in Arabidopsis (Jelenska et al., 2007).

Studies of nuclear architecture and dynamics have been emerging as hot topics in plant science (Cheung and Reddy, 2012). The nucleoplasm is organized into non-membrane-bound subdomains and bodies, which play diverse roles in DNA replication, gene expression, gene silencing, proteolysis and RNA splicing (Mao et al., 2011). The proteins that control these nuclear processes are highly organized into discrete bodies within the nucleus (Spector and Lamond, 2011). In plants, the best-investigated nuclear bodies are the nucleolus, Cajal bodies, dicing bodies (D-bodies), photobodies (photo-morphogenesis) and nuclear speckles, which contain Ser/Arg-rich (SR) proteins and numerous other proteins, and play critical roles in pre-mRNA or alternative splicing (reviewed in Reddy et al., 2012). The SR-containing nuclear speckles are located in the interchromatin space and are considered to be storage sites for splicing factors (Fang and Spector, 2007; Spector and Lamond, 2011). SR proteins contain RNA recognition motifs (RRMs) at the N terminus and an Arg/Ser-rich (SR) domain at the C terminus.

Even though DnaJ proteins have been studied in great detail in the model plant Arabidopsis, their roles in crop plants such as Glycine max (soybean) are still largely uninvestigated. Here, we showed that transient overexpression of GmHSP40.1, a soybean type-III nuclear body-localized DnaJ protein, in Nicotiana benthamiana results in hypersensitive response (HR)-like cell death. Induction of cell death was dependent on nuclear localization and on the MAPKKKα- and wounding-inducible protein kinase (WIPK)-mediated signaling pathway. GmHSP40.1 co-localized with SERRATE (SE), a protein involved in pri-miRNA and/or pre-mRNA splicing, suggesting that cell death induced by overexpressing GmHSP40.1 might be a consequence of altered pre-mRNA splicing or pri-miRNA processing. Consistent with its role as a positive regulator of HR-like cell death, silencing GmHSP40.1 enhanced the susceptibility of soybean plants to Soybean mosaic virus (SMV), and its overexpression in transgenic Arabidopsis resulted in a cell death and/or early senescence phenotype, accompanied by activated defense responses. Collectively, our studies reveal a new role for a nuclear-localized DnaJ domain-containing protein in plant cell death and disease resistance.


Transient overexpression of DnaJ domain-containing GmHSP40.1 from soybean leads to HR-like cell death in N. benthamiana

In a screen for candidate proteins stimulating cell death, full-length cDNAs of soybean genes in MPK signaling pathways, and other biotic and abiotic resistance pathways, were cloned into a binary T-DNA vector for transient expression via Agrobacterium infiltration (agroinfiltration), which is a commonly used approach to study proteins with functions in cell death or HR (Ekengren et al., 2003). It usually takes 1.5 days for proteins to reach maximal expression using agroinfiltration, and HR can be visible after 2 days post-infiltration (dpi). Overexpression of a DnaJ domain-containing GmHSP40.1 resulted in HR-like cell death in the infiltrated leaf area. The first signs of cell death were observed as water soaking in the infiltrated areas at 2 dpi, and the HR cell death became apparent at 3 dpi (Figure 1). No HR was observed when the GFP negative control and two MAP kinases were transiently overexpressed (Figure 1), indicating that GmHSP40.1 positively regulated cell death in this assay. Phylogenetic analysis indicated that GmHSP40.1 is a type-III DnaJ protein (Rajan and D'Silva, 2009), because its amino acid sequence lacks both the HPD and zinc-finger (CxxCxGxG) domains. Because a role for a plant HSP40.1 in cell death has not been reported previously, we characterized this protein further.

Figure 1.

Transient overexpression of GmHSP40.1 in Nicotiana benthamiana leaves causes hypersensitive response (HR)-like cell death. Agrobacterium tumefaciens (strain GV2260) carrying 35S-GFP, 35S-GmMKK6 (Glyma10g15850) and 35S-GmMKK2a (Glyma17g06020), or 35S-GmHSP40.1 were diluted to OD600 = 0.5 and infiltrated into N. benthamiana leaves. This representative leaf was photographed at 3 days post infiltration.

GmHSP40.1 is a nuclear-localized protein

In addition to the DnaJ domain, a nuclear localization signal (NLS) with a sequence of RRRKRRKRR was identified at amino acids 325–333 (Figure S1). To investigate the subcellular localization of GmHSP40.1, the coding sequence was fused with GFP to make 35S::GFP-GmHSP40.1, and the resulting plasmid was delivered into onion cells via biolistic bombardment. Consistent with the presence of the predicted NLS, strong GFP fluorescence was detected in the nucleus of onion cells (Figure 2a), indicating that GFP-GmHSP40.1 was nuclear localized. As a control, free RFP was co-expressed and RFP signal was present both in the cytoplasm and nucleus (Figure 2a). Interestingly, under higher magnification GFP-GmHSP40.1 was seen in distinctive bodies inside the nucleus (Figure 2b). Similar nuclear bodies were also observed when GFP-GmHSP40.1 was transiently expressed in N. benthamiana leaf cells. The size and number of the nuclear bodies varied from cell to cell (Figure 2b). Each nucleus contained between three and 25 nuclear bodies, ranging in size from 0.1 to 1 μm, but the identity of the nuclear bodies was not known.

Figure 2.

GmHSP40.1 is localized in nuclear bodies.(a) GFP-GmHSP40.1 and free DsRED constructs driven by the 35S promoter were co-bombarded into onion epidermal cells. The left panel shows the transient expression of GFP-GmHSP40.1; the second panel shows the transient expression of free RFP; the third panel shows the bright-field image of bombarded onion cells; and the right panel shows the merged image. Scale bar: 100 μm.(b, c) Agrobacterium tumefaciens (strain GV2260) carrying GFP-GmHSP40.1 was infiltrated into Nicotiana benthamiana leaves. Close-up images showing that GFP-GmHSP40.1 is present both in nucleoli (red arrows in B) and nuclear speckles (white arrows). Scale bars: 20 μm.

Nuclear localization is required for cell death induced by overexpression of GmHSP40.1

To test whether nuclear localization of GmHSP40.1 is required for the induction of cell death, a nuclear exclusion signal (NES; Shen et al., 2007) was fused to its C terminus to generate GmHSP40.1-NES, and the resulting construct was introduced into N. benthamiana leaves via agroinfiltration. In contrast to wild-type GmHSP40.1 (Figure 1), GmHSP40.1-NES did not trigger cell death in the infiltrated area (Figure 3d), indicating that nuclear localization of GmHSP40.1 was necessary to trigger cell death, and therefore was critical for its function. To confirm that the NES prevented GmHSP40.1 from entering the nucleus, the GFP-GmHSP40.1-NES construct was made and transiently expressed in the leaves of Nbenthamiana via agroinfiltration. The NES drastically reduced the nuclear localization of GFP-GmHSP40.1-NES (Figure 3b). In most cells, GFP fluorescence was only observed in the cytoplasm, and the number of cells in which GFP-GmHSP40.1-NES was targeted to the nuclei was only 12% of the number of cells in which GFP-GmHSP40.1 was targeted to the nuclei (compare panels a and b in Figure 3).

Figure 3.

Nuclear localization is required for the cell death-inducing function of GmHSP40.1.Compared with the exclusive nuclear localization of GFP-GmHSP40.1 (a), fusing an NES signal peptide to GFP-GmHSP40.1 (GFP-GmHSP40.1-NES) drastically reduced the nuclear localization to only 12% of the cells (b), whereas deleting the NLS signal motif within GmHSP40.1 completely abolished its nuclear localization (GFP-GmHSP40.1-ΔNLS) (c). Agrobacterium tumefaciens solutions (OD600 = 0.5) carrying 35S-GFP, 35S-GmHSP40.1, GmHSP40.1-NES and GmHSP40.1-ΔNLS were infiltrated into Nicotiana benthamiana leaves to test their abilities to induce cell death (d). This representative leaf was photographed at 3 days post infiltration.

To further confirm the importance of nuclear localization in triggering cell death, the predicted NLS, RRRKRRKRR (aa 325–333), was deleted to generate the GmHSP40.1-ΔNLS construct. GFP-GmHSP40.1-ΔNLS was exclusively localized in the cytoplasm (Figure 3c), demonstrating that GmHSP40.1 was directed to the nucleus by the NLS. The expression of GmHSP40.1-ΔNLS did not lead to cell death in N. benthamiana leaves (Figure 3d), confirming that nuclear localization of GmHSP40.1 is required for the induction of cell death.

GFP-GmHSP40.1 co-localizes with HcRed-SE

Only a few plant proteins have been reported to reside in nuclear bodies or speckles, including DCL1, HYL1, Coilins, SE and SR33 (Fang and Spector, 2007; Reddy, 2004; Reddy et al., 2012). To understand the nature (components and function) of the nuclear bodies containing GmHSP40.1, RFP-GmHSP40.1 was transiently co-expressed with DCL1-YFP, HYL1-YFP and Coilin-YFP. RFP-GmHSP40.1 was not co-localized with DCL1-YFP, HYL1-YFP or Coilin-YFP (Figure S2). To our surprise, when transiently co-expressed with HcRED-SE, GFP-GmHSP40.1 nuclear bodies were either partially or completely co-localized with HcRED-SE nuclear bodies in the majority of co-expressing cells (over 80%; Figure 4, middle and lower panel). In only 10–20% of the co-expressing cells, the nuclear bodies formed by GmHSP40.1 and HcRED-SE were not co-localized (Figure 4, upper panel). These results suggest that the components of these dynamic nuclear bodies are not homogenous and they change over time, or GFP-HSP40.1 can incorporate into HcRED-SE bodies in the majority of cells. SERRATE, a zinc-finger protein, regulates pre-mRNA splicing and pri-miRNA processing (Laubinger et al., 2008; Raczynska et al., 2010). It has been shown that HcRED-SE co-localizes with SR33-YFP (Fang and Spector, 2007), which is also involved in pre-mRNA splicing. These results suggest that GmHSP40.1 may play a role in pre-mRNA splicing and mRNA maturation, or in small RNA biogenesis.

Figure 4.

GmHSP40.1 is co-localized with HcRed-SE.Agrobacterium tumefaciens (strain GV2260) carrying GFP-GmHSP40.1 and HcRed-SE was co-infiltrated into Nicotiana benthamiana leaves. At 2 days post infiltration, the infiltrated area was cut into 0.5-cm2 pieces that were examined by fluorescence microscopy. In some cells, GmHSP40.1 (green) is not co-localized with HcRed-SE (upper panel); however, in the majority of cells, GFP-GmHSP40.1 is either partially co-localized with HcRed-SE (middle panel) or almost completely co-localized with HcRed-SE (lower panel). The arrows indicate the signals that are not co-localized between GFP-HSP40.1 and HcRed-SE. Scale bar: 20 μm.

MAPKKKα, WIPK and SIPK are required for cell death induced by the overexpression of GmHSP40.1

Mitogen-activated protein (MAP) kinase cascades are required for cell death triggered by overexpressing LeMAPKKKα (del Pozo et al., 2004), and LeMAPKKKα is required for cell death caused by silencing Adi3 in Solanum lycopersicum (tomato; Devarenne et al., 2006). To test whether MAP kinase cascades are required for the cell death triggered by overexpressing GmHSP40.1, a Tobacco rattle virus (TRV) vector was used to silence NbMAPKKKα, NbMEK1, NbMEK2 or NbSIPKK, and NbSIPK, NbWIPK, NbNTF4 or NbNTF6 in N. benthamiana. Plants infected with the TRV empty vector (TRV2:0) served as a negative control. GmHSP40.1 was transiently expressed in the silenced leaves via agroinfiltration. As expected, cell death was observed in infiltrated leaf areas expressing GmHSP40.1 in TRV2:0-treated plants (Figure 5). Silencing NbMEK1, NbMEK2, NbSIPPK1, NbNTF4 or NbNTF6 had no effect on GmHSP40.1-induced cell death (Figure 5). However, silencing NbMAPKKKα and NbWIPK completely suppressed cell death induced by GmHSP40.1 (Figure 5). Silencing NbSIPK delayed GmHSP40.1-induced cell death (Figure 5). Together, our results indicate that NbMAPKKKα-, NbWIPK- and NbSIPK-mediated signaling pathways are required for GmHSP40.1-induced cell death.

Figure 5.

Cell death induced by transient overexpression of GmHSP40.1 is dependent upon WIPK and MAPKKKα.The genes encoding the MAP kinases, NbWIPK, NbSIPK and NbNTF6, MAP kinase kinases NbMEK1, NbMEK2 and NbSIPKK, and MAP kinase kinase kinase, NbMAPKKKα, were silenced via TRV-mediated virus-induced gene silencing (VIGS) in Nicotiana benthamiana. At 18 days post infiltration, Agrobacterium tumefaciens (strain GV2260) (OD600 = 0.5) carrying 35S-GmHSP40.1 was infiltrated into these N. benthamiana plants silenced for various genes involved in the MAP kinase cascades. These representative leaves were photographed at 3 days post 35S-GmHSP40.1 infiltration.

Silencing GmHSP40.1 results in the enhanced susceptibility of soybean plants to SMV

Our results so far suggested that GmHSP40.1 positively regulates plant defense, and we expected that silencing its expression would compromise plant defenses. Such a role would be consistent with HSP90 and HSP70, which play important roles in defense responses, and knocking down their expression compromises disease resistance (Hubert et al., 2003; Lu et al., 2003; Takahashi et al., 2003; Liu et al., 2004). To test this possibility, we silenced GmHSP40.1 using the Bean pod mottle virus virus-induced gene silencing (BPMV-VIGS) vector (Zhang et al., 2010). Silencing of GmHSP40.1 was confirmed by RT-PCR (Figure 6a). GmHSP40.1-silenced and empty vector control plants were inoculated with an infectious clone of SMV that expresses the GUS enzyme (SMV-N-GUS) (Wang et al., 2006). Four individual leaves from four independent vector control plants or GmHSP40.1-silenced plants were used in this assay. The SMV-N-GUS infection was visualized by GUS staining at 3 dpi (Figure 6b). The diameters of GUS foci were measured on the leaves of both GmHSP40.1-silenced plants and vector control plants. The sizes of GUS foci were significantly increased (Student's t-test, P < 0.01) by ~30% on GmHSP40.1-silenced leaves, compared with vector control leaves, demonstrating that silencing GmHSP40.1 enhances susceptibility to SMV infection (Figure 6c).

Figure 6.

SMV-N-GUS infectivity is enhanced in GmHSP40.1-silenced soybean leaves.At 20 days post inoculation (dpi) with BPMV constructs, SMV-N-GUS was biolistically delivered into detached leaves of either BPMV-0 or BPMV-GmHSP40.1 plants. The diameters of GUS foci were determined at 3 dpi with SMV-N-GUS.(a) RT-PCR showing that the transcript level of GmHSP40.1 was decreased in GmHSP40.1-silenced plant (S) relative to empty BPMV vector control plant (V) (the colors of image were inverted).(b) Comparison of the infection foci of SMV-N-GUS on the leaves of empty vector control (BPMV-0) and GmHSP40.1-silenced plants (BPMV-GmHSP40.1).(c) Comparison of the size of SMV-N-GUS foci on the leaves of BPMV-0 and BPMV-GmMPK40.1 plants. The data shown in (c) are mean values of four individual leaves from four different plants (four leaves in total). The error bars in (c) represent standard deviations for four independent leaves. At least 30 GUS foci from each of four independent leaves were measured for (c). **P < 0.01, Student's t-test.

Cell death and accelerated senescence were observed in transgenic Arabidopsis plants overexpressing GmHSP40.1

To further confirm a cell death-inducing function for GmHSP40.1, transgenic Arabidopsis plants were generated that ectopically overexpressed 35S-GmHSP40.1. The authenticity of the transgenic plants was confirmed by genomic PCR and RT-PCR (Figure 7d). Among 20 independent transgenic plants, four had cell-death or early-senescence phenotypes (Figure 7a) that became more severe at later developmental stages (Figure 7b). At 50 days after germination the leaves of Col-0 remained green (Figure 7e), whereas most leaves of the transgenic 35S-GmHSP40.1 plants were almost dried out (Figure 7d). This result is consistent with the transient expression of GmHSP40.1 in N. benthamiana (Figure 1).

Figure 7.

Transgenic Arabidopsis lines overexpressing GmHSP40.1 display chlorosis and cell-death phenotypes with activated defense responses.Chlorosis (a) and cell death (b) on 50-day-old transgenic lines overexpressing GmHSP40.1 driven by the 35S promoter. The Col-0 plant at the same developmental stage was used as a control (c). Enlarged images of leaves from the plants shown in (b) and (c) are shown in (d) and (e), respectively. The expression of AtPR1 and AtPR5 was induced in the transgenic plants with cell death (f). The authenticity of transgenic plants was confirmed by the expression of GmHSP40.1. AtUBQ5 used as a control. The colors of RT-PCR images were inverted.

To test whether the cell death or early senescence phenotype correlated with an activated defense response, PR gene expression was compared between Col-0 and transgenic 35S-GmHSP40.1 plants. Expression of both AtPR1 and AtPR5 mRNAs was greatly induced in transgenic 35S-GmHSP40.1 plants (Figure 7f), indicating that a defense response is activated in these plants.


Nuclear-localization of GmHSP40.1 is required for its function

DnaJ proteins reside in the cytosol, nucleus, endosomes, mitochondria, chloroplast, ER and ribosomes (Qiu et al., 2006; Rajan and D'Silva, 2009). There are 24 putative nuclear-localized DnaJ proteins in Arabidopsis (Rajan and D'Silva, 2009), but only one of them has been functionally characterized (Park et al., 2011). Here, we describe a role for nuclear-localized GmHPS40.1 as a positive regulator of plant defense responses. We used two independent strategies to prevent GmHSP40.1 from entering the nucleus by deleting its NLS or fusing the coding sequence to an NES. These data demonstrated that the nuclear localization of this DnaJ protein is necessary for its cell death-inducing function (Figure 3). It is possible that deleting the NLS or fusing the NES may alter the protein structure, and thus affect its function; however, it seems unlikely that both approaches would have deleterious effects. Addition of the NES is a widely used approach to prevent fusion proteins from entering the nucleus. In studies related to plant defense, fusion of the NES to R proteins, MLA and N was a key piece of evidence used to demonstrate that their nuclear localization is necessary for the induction of defense (Burch-Smith et al., 2007; Shen et al., 2007). Fusion of the NES to GmHPSP40.1 did not completely abolish nuclear localization, as there were still 12% of cells with nuclear localization (Figure 3). This observation suggests that an equilibrium between targeting to the nucleus mediated by the NLS and exclusion from the nucleus mediated by the NES exists, and that in a majority of cells the NES was dominant over the NLS. The fact that GmHSP40.1-NES did not induce cell death (Figure 3d) suggests that there is a threshold level for this nuclear-localized DnaJ protein in inducing a visible cell death.

MAPKKKα and WIPK are required for GmHSP40.1-triggered cell death

Mitogen-activated protein kinase cascades are often involved in activating or suppressing cell death (Petersen et al., 2000; Liu et al., 2004, 2011; and Popescu et al., 2009). MAPKKKα has been implicated in Pto-mediated cell death, as silencing MAPKKKα suppressed cell death triggered by Pto/AvrPto or an activated PtoY207D mutant (del Pozo et al., 2004). Overexpressing MAPKKKα can result in pathogen-independent cell death (del Pozo et al., 2004). By overexpressing MAPKKKα in leaves in which various MAPKK and MAPK genes were silenced by VIGS, del Pozo et al. (2004) identified three distinct MAP kinase signaling modules that mediate cell death. MAPKKKα acts upstream of MEK2 (MAPKK)/SIPK (MAPK), but the MAPKKKs that act upstream of the MEK2-WIPK and MEK1-NTF6 cascades remain unidentified. Interestingly, cell death induced by the expression of NtMEK2DD was fully suppressed in leaves silenced for SIPK, WIPK or MEK1, and partially in leaves silenced for NTF6 (del Pozo et al., 2004). Silencing MAPKKKα also compromises cell death triggered by silencing Adi3 in tomato (Devarenne et al., 2006). Unlike what has been observed for Pto/AvrPto-triggered cell death, we found that MAPKKKα and WIPK, but not MEK1 or MEK2, are required for HSP40-activated cell death, as WIPK acts downstream of MEK2 in a MAPKKKα-independent module in Pto/AvrPto-triggered cell death. These results indicate that hierarchical and specific MAPK signaling modules, culminating in cell death, are engaged when GmHSP40.1 is overexpressed in N. benthamiana; however, the MAPKK that functions as intermediary between MAPKKKα and WIPK remains to be identified. Nonetheless, our results show that cell death triggered by GmHSP40.1 overexpression is achieved through the activation of an MAP kinase module containing MAPKKKα and WIPK.

Roles of DnaJ-HSP40 in disease resistance

DnaJ proteins have previously been reported to interact with viral movement or coat protein, and serve as host susceptibility factors (Hofius et al., 2007). Silencing of NtMPIP1 or overexpression of a J domain-deficient version of NtCPIP1 significantly inhibited the spread of TMV and PVX (Shimizu et al., 2009). It was proposed that CPIPs act as important susceptibility factors during viral infection, possibly by recruiting HSP70 chaperones for viral assembly and/or cellular spread. Unexpectedly, we found that instead of inhibiting the spread of SMV, silencing GmHSP40.1 significantly enhanced SMV-GUS infectivity (Figure 6), suggesting that different HSP40 homologs may play opposite roles in viral infection. These different roles are presumably a result of the subfunctionalization of family members, including differences in subcellular localization. Similarly, previous results by others show that different DnaJ proteins can play opposing roles in flowering time. Mutation of a member of the DnaJ III family, which is localized both in the cytosol and nucleus, delays flowering in Arabidopsis (Shen et al., 2011), but a null mutant of another nuclear-localized DnaJ family member flowers early (Park et al., 2011). These contrasting effects on phenotype indicate that different DnaJ proteins play distinct roles in various processes.

Interestingly, opposing roles of host cytosolic HSP70s, the chaperones of DnaJ-HSP40, on pathogen infection have also been reported previously. Some cytosolic HSP70s facilitate proper folding of viral proteins and promote viral infectivity (replication etc.) (Nagy et al., 2011). However, HSP70s are critical for basal resistance against bacteria in Arabidopsis (Jelenska et al., 2010), INF1-mediated HR and non-host resistance to Pseudomonas cichorii in N. benthamiana (Kanzaki et al., 2003). Interestingly, HopI1 is an effector that is targeted to the chloroplast, and it contains a type-III J domain. HopI1 suppresses SA accumulation and related defense responses in Arabidopsis (Jelenska et al., 2007). HopI1 directly binds cytosolic HSP70 through its C-terminal J domain, and stimulates HSP70 ATP hydrolysis activity in vitro (Jelenska et al., 2010). It has been hypothesized that HopI1 subverts defense-promoting activity/activities) of HSP70 through inducing and recruiting cytosolic HSP70 to chloroplasts (Jelenska et al., 2010). Besides HSP70, DnaJ/HSP40 proteins can also regulate other chaperones, such as the 90-kDa heat-shock protein HSP90. HSP70 and HSP90 cooperate in the folding of many substrates in the eukaryotic cytosol. The DnaJ/HSP40 protein TPR2 (DnaJC7) has been suggested to mediate the retrograde transfer of substrates from HSP90 onto HSP70 (Brychzy et al., 2003). Cytosolic HSP90 is a major component in R gene-mediated resistance, and knocking out or knocking down HSP90 in Arabidopsis, N. benthamiana or soybean compromises RPM1, RPS2, Rx, N and Rsv1 gene-mediated resistance (Hubert et al., 2003; Lu et al., 2003; Takahashi et al., 2003; Liu et al., 2004; Zhang et al., 2012). In our case, silencing GmHSP40.1 did not disrupt Rsv1-mediated extreme resistance against SMV (Figure S3). Whereas GmHSP40.1 does not seem to have a role in R gene-mediated resistance to SMV, its silencing did result in enhanced susceptibility. This result indicated that GmHSP40.1 plays a role in basal defense against SMV. In addition to serving as co-chaperones, DnaJ in bacteria and certain other members of the DnaJ/Hsp40 family can be chaperones by themselves through binding to certain unfolded proteins and nascent peptide chains (Hendershot et al., 1996), suggesting that DnaJ/Hsp40 may function independently of HSP70 or HSP90. Because of its nuclear localization, it is unlikely that silencing GmHSP40.1 has a direct effect on cytosolic HSP70 and HSP90 functions, which would be expected to participate in SMV replication and movement. The enhanced SMV-GUS infection observed on GmHSP40.1-silenced plants is likely a result of direct or indirect effects of GmHSP40.1 being absent from the nucleus. Such effects could involve altered expression of host genes involved in defense against viral and other pathogens.

RNA splicing, R protein activation and cell death

The observation that GmHSP40.1 co-localized with HcRed-SE (Figure 3), which itself co-localizes with SR33 (Fang and Spector, 2007), in nuclear bodies opens up some interesting hypotheses on how GmHSP40.1 might influence the expression of host genes involved in pathogen defense. SR proteins, which contain an arginine/serine rich (SR) domain, are highly conserved families of nuclear proteins that play important roles in pre-mRNA splicing and alternative splicing (Lopato et al., 1996, 1999; Reddy, 2004; Barta et al., 2008; Long and Caceres, 2009). Alternative splicing is a common feature for plant R genes such as the tobacco N gene (Whitham et al., 1994), barley Mla13 (Halterman et al., 2003), Arabidopsis SNC1 (Yi and Richards, 2007) and RPS4 (Zhang and Gassmann, 2003, 2007). Alternative splicing is required for the function of some R genes, including N (Dinesh-Kumar and Baker, 2000), RPS4 (Zhang and Gassmann, 2003, 2007) and SNC1 (Xu et al., 2011). Multiple transcript variants (TVs) and two TVs have been identified for RPS4 (Zhang and Gassmann, 2003, 2007) and N (Dinesh-Kumar and Baker, 2000), respectively. A lack of TVs results in compromised N- and RPS4-mediated resistance (Dinesh-Kumar and Baker, 2000; Zhang and Gassmann, 2003, 2007). The importance of alternative splicing for R gene function is further supported by a recent study showing that loss of a Transportin-SR (TRN-SR) (mos14-1), a member of the importin-β superfamily that functions as the nuclear import receptor for SR proteins, results in altered splicing patterns of SNC1 and RPS4, and compromised resistance mediated by snc1 and RPS4 (Xu et al., 2011). The fact that GmHSP40.1 is co-localized with SE, which is co-localized with SR33 (Fang and Spector, 2007), suggests that it might play a role in pre-RNA splicing or alternative splicing. It will be interesting to test whether the cell death caused by overexpression of GmHSP40.1 is associated with R gene mis-activation or a response resulting from altered general RNA splicing. The involvement of pre-mRNA splicing in cell death was also shown by a recent study of Oryza sativa (rice) carrying a mutation in the putative splicing factor 3b subunit 3 (SF3b3), which displays a lesion mimic symptom and enhanced disease resistance against rice blast (Chen et al., 2012).

GmHSP40.1 co-localizes with HcRED-SE (Figure 3). The SE protein is an essential component for both pri-miRNA and pre-mRNA processing (Laubinger et al., 2008). SE presumably works as a scaffold-like protein capable of binding both protein and RNA to guide the positioning of the miRNA precursor towards the DCL1 catalytic site within the miRNA processing machinery in plants (Laubinger et al., 2008; Machida et al., 2011). The increased SMV-GUS foci observed in GmHSP40.1-silenced plants could be caused by compromised SE function in miRNA biogenesis, given that the biogenesis of miRNAs play important roles in antiviral defense (Moissiard and Voinnet, 2006). This is another interesting possibility that can be tested in future experiments.

Experimental Procedures

Plant material

Soybean cultivar Williams 82, N. benthamiana and Arabidopsis thaliana ecotype Col-0 were used in this study. Soybean plants were maintained in a glasshouse or growth chamber at 22°C with a photoperiod of 16 h. Arabidopsis or N. benthamiana plants were grown in a growth room at 20°C during the dark and 22°C during the light, with a photoperiod of 16 h.

Constructs and agroinfiltration

The full-length coding sequence of GmHSP40.1 (Glyma15g06290) was amplified by RT-PCR from total RNA extracted from Williams 82 soybean plants using the following pair of primers: GmHSP40.1-F, 5′-CACCATGGATGGTCACGGAGGAG-3′; GmHSP40.1-R, 5′-ATGACCCTTAACCTTATCATCTGCAA-3′. The GmHSP40.1 coding sequence fused to the NES at the C terminus was amplified using the following pair of primers: GmHSP40.1-F, 5′-CACCATGGATGGTCACGGAGGAG-3′; GmHSP40.1-NES-R, 5′-CTATTTGTTAATATCTAGACCAGCCAGCTTCAGGGCCAGTTCGTTATGACCCTTAACCTTATCATCTGCAA-3′. The underlined nucleotides indicate the NES sequence (Shen et al., 2007).

The GmHSP40.1 coding sequence with its NLS deleted was generated using a fusion PCR strategy. The fragment upstream of the NLS was amplified using the following pair of primers: GmHSP40.1-F1, 5′-CACCATGGATGGTCACGGAGGAG-3′; GmHSP40.1Fu-R2, 5′-CATCACCACCAGCATTGCTCGTGGTCCCCTTCTGA-3′. The fragment downstream of the NLS was amplified using the following pair of primers: GmHSP40.1Fu-F2, 5′-TCAGAAGGGGACCACGAGCAATGCTGGTGGTGATGT-3′; GmHSP40.1-R1, 5′-ATGACCCTTAACCTTATCATCTGCAA-3′. The two PCR products lack the NLS, but share 35 bp of overlapping complementary sequence. Subsequently, the full-length coding sequence lacking the NLS was amplified using GmHSP40.1-F1 and GmHSP40.1-R1 as forward and reverse primers, and the two PCR products mixed at a 1:1 molar ratio as a template. The amplified coding sequences described above were cloned into the pENTR/D entry vector (Invitrogen, http://www.invitrogen.com) to generate pENTR/D/GmHSP40.1, pENTR/D/GmHSP40.1-NES and pENTR/D/GmHSP40.1-ΔNLS. These inserts were subsequently recombined into binary destination vectors pB7WG2,0, and pB7WGF2,0 (Karimi et al., 2002) via an LR reaction to generate 35S-GmHSP40.1, 35S-GFP-GmHSP40.1, 35S-GmHSP40.1-NES, 35S-GFP-GmHSP40.1-NES, 35S-GmHSP40.1-ΔNLS and 35S-GFP-GmHSP40.1-ΔNLS. Agrobacterium tumefaciens (strain GV2260) containing the binary vectors were grown and infiltrated into leaves as described by Ding et al. (2004).

Subcellular localization of GmHSP40.1

35S-GFP-GmHSP40.1, 35S-GFP-GmHSP40.1-NES and 35S-GFP-GmHSP40.1-ΔNLS fusion constructs were either bombarded into onion epidermal cells (Biolistic PDS-1000/He system (Bio-Rad Laboratories, Hercules, CA, USA), as described by Zhang et al. (2009), or transiently expressed in N. benthamiana leaves via agroinfiltration, as described by Liu et al. (2005). Images were captured with an inverted Axiophot microscope (Zeiss, Jena, Germany) equipped with a digital camera (Diagnostic Instruments, Sterling Heights, MI, USA).

TRV-mediated VIGS in N. benthamiana

For the VIGS assay, the pTRV1 and pTRV2 silencing vectors were provided by S.P. Dinesh-Kumar (Liu et al., 2002). The TRV constructs used for silencing various kinase genes (pTRV1, pTRV2-WIPK, pTRV2-SIPK, pTRV2-MEK1, pTRV2-MEK2, pTRV2-MAPKKKα, pTRV2-N) in this study were provided by Gregory Martin, and were described (Ekengren et al., 2003). Agrobacterium growth and infiltration were performed as described by Ding et al. (2004).

BPMV-mediated VIGS in soybean

BPMV VIGS vectors, pBPMV IA-R1M and pBPMV-IA-D35 were used in this study (Zhang et al., 2010). pBPMV-IA-D35 is a derivative of pBPMV-IA-R2 containing BamHI and KpnI restriction sites between the cistrons encoding movement protein and the large coat protein subunit. A fragment of the GmHSP40.1 coding sequence was amplified using the forward (5′-AAGGGATCCACAATTGAGAGGCCTAGAAGACG-3′) and the reverse (5′-TTGGGTACCCCTTATCATCTGCAACAACATTGAGA-3′) primers. The underlined sequences are BamHI and KpnI restriction sites, respectively. The nucleotide set in bold indicates the extra nucleotide in the reverse primer that is needed to maintain the reading frame. The BPMV constructs were inoculated onto soybean by biolistic particle bombardment using a Biolistic PDS-1000/He system (Bio-Rad Laboratories, http://www.bio-rad.com).

SMV-N-GUS inoculation, GUS staining and GUS foci measurements

At 18 dpi with BPMV empty vector (BPMV-0) or BPMV-GmHSP40.1 constructs, the second fully expanded soybean trifoliate leaves counting from the top were detached and biolistically inoculated with SMV-N-GUS (Wang et al., 2006; Zhang et al., 2009). Following SMV-N-GUS inoculation, the detached leaves were incubated on moist filter paper in Petri dishes and kept on a lit growth shelf for 3 days before GUS staining. GUS staining was performed as described by Jefferson et al. (1987). Photos of the leaves with GUS foci were taken under a stereomicroscope (Olympus SZH10, http://www.olympus-global.com). The diameters of GUS foci were measured using the Soft Image System (SIS) Analysis (ia package; Olympus).

Generation of Arabidopsis transgenic lines

Binary vectors harboring the cassettes of CaMV35S:GmHSP40.1 were introduced into A. tumefaciens (strain GV2260) using standard protocols. Plants were transformed as described by Clough and Bent (1998). Transformants were selected by spraying herbicide (Finale; Bayer CropScience, http://www.cropscience.bayer.com) and confirmed by genomic PCR and RT-PCR.

RNA isolation and RT-PCR

RNA isolation and RT-PCR were performed as described elsewhere (Liu et al., 2005). The RNA samples were treated with RNAse free DNaseI according to the manufacturer's instructions (Invitrogen). The primers used in this study were: AtUBQ5-F, 5′-GTGGTGCTAAGAAGAGGAAGA-3′, and AtUBQ5-R, 5′-TCAAGCTTCAACTCCTTCTTT-3′; AtPR1-F, 5′-TAGCCCACAAGATTATCTAAGG-3′, and AtPR1-R, 5′-CTCGTTCACATAATTCCCAC-3′; AtPR5-F, 5′-ATGGCAAATATCTCCAGTATTCAC-3′, and AtPR5-R, 5′-GAAAATCCTCGAGTAGTCCGT-3′; GmHSP40.1-F, 5′-CACCATGGATGGTCACGGAGGAG-3′; and GmHSP40.1-R, 5′-ATGACCCTTAACCTTATCATCTGCAA-3′.


The authors gratefully acknowledge: Greg Martin for N. benthamiana silencing constructs; S.P. Dinesh-Kumar for the TRV silencing vector; Yuda Fang for HcRED-SE, YFP-Coilin1, YFP-DCL1 and YFP-HYL1; and Jaime Dittman, Al Eggenberger, Wenli Qiu, Dali Wang, Jack Horner and Randall Den Adel for their excellent technical assistance. This work was supported by the NSF Plant Genome Research Program (award number 0820642), the Iowa Soybean Association, the United Soybean Board, the North Central Soybean Research Program, Hatch Act and State of Iowa Funds and by the Natural Science Foundation of Zhejiang Province (award number LY12C14001) and Zhejiang Normal University to JZL. This is a journal article of the Iowa Agriculture and Home Economics Experiment Station, Ames, IA, project number 3708.