The Arabidopsis extracellular UNUSUAL SERINE PROTEASE INHIBITOR functions in resistance to necrotrophic fungi and insect herbivory


(fax (765) 4940363; e-mail


Protease inhibitors (PIs) function in the precise regulation of proteases, and are thus involved in diverse biological processes in many organisms. Here, we studied the functions of the Arabidopsis UNUSUAL SERINE PROTEASE INHIBITOR (UPI) gene, which encodes an 8.8 kDa protein of atypical sequence relative to other PIs. Plants harboring a loss-of-function UPI allele displayed enhanced susceptibility to the necrotrophic fungi Botrytis cinerea and Alternaria brassicicola as well as the generalist herbivore Trichoplusia ni. Further, ectopic expression conferred increased resistance to B. cinerea and T. ni. In contrast, the mutant has wild-type responses to virulent, avirulent and non-pathogenic strains of Pseudomonas syringae, thus limiting the defense function of UPI to necrotrophic fungal infection and insect herbivory. Expression of UPI is significantly induced by jasmonate, salicylic acid and abscisic acid, but is repressed by ethylene, indicating complex phytohormone regulation of UPI expression. The upi mutant also shows significantly delayed flowering, associated with decreased SOC1 expression and elevated levels of MAF1, two regulators of floral transition. Recombinant UPI strongly inhibits the serine protease chymotrypsin but also weakly blocks the cysteine protease papain. Interestingly, jasmonate induces intra- and extracellular UPI accumulation, suggesting a possible role in intercellular or extracellular functions. Overall, our results show that UPI is a dual-specificity PI that functions in plant growth and defense, probably through the regulation of endogenous proteases and/or those of biotic invaders.


Protease inhibitors (PIs) and their target proteases regulate many diverse plant biological processes. In response to developmental and environmental cues, proteases irreversibly cleave target proteins for proteolytic processing and degradation, thereby regulating a wide range of cellular responses (Koiwa et al., 1997; van der Hoorn, 2008). Plant proteases have been implicated in the regulation of defense, embryogenesis, cell fate, cuticle formation and general cellular housekeeping (van der Hoorn, 2008). Thus, plants must stringently govern protease activity in order to achieve cellular homeostasis. PIs serve as one mechanism of protease regulation by directly binding to catalytic sites and blocking activity (Laskowski and Kato, 1980). The Arabidopsis genome encodes more than 800 proteases and an approximately equivalent number of PIs (Rawlings et al., 2006; Jongsma and Beekwilder, 2008). Although largely studied with respect to their anti-microbial activities, PIs function in programmed cell death, seed maturation, tracheary formation and many other plant processes (Jongsma and Beekwilder, 2008).

Many proteases and PIs are associated with pathogen virulence and plant defense against pathogens and insects. To obtain nutrition, insects utilize various proteases to digest host tissues (Botella et al., 1996). In turn, plants have co-evolved the counter defense of accumulating PIs both at the site of feeding and systemically. Upon ingestion, host PIs block gut proteases, decreasing the rate of insect digestion and thereby limiting their nutritional absorption and growth despite feeding (Botella et al., 1996). Plant pathogens also use proteases to aid in colonization and active suppression of host defense responses, thus many PIs serve dual roles as anti-feedants as well as anti-microbial agents. A pearl millet (Pennisetum glaucum) cysteine PI acts as an anti-feedant and possesses anti-fungal activity against Trichoderma, Fusarium and Alternaria species (Joshi et al., 1999). The fungal pathogen Botrytis cinerea secretes proteases with trypsin and chymotrypsin-like activity that are probably involved in degradation of host cell walls (Brown and Adikaram, 1983; Movahedi and Heale, 1990). Incubation of tomato (Solanum lycopersicum) cell-wall proteins with secreted B. cinerea proteases inhibited activity by more than 70%, suggesting a role for host PIs in defense against such broad-host pathogens (Brown and Adikaram, 1983). During tomato development, PIs can be detected throughout ripening, with a reduction upon fruit maturation. These PIs have a dual function, allowing slow metabolic release of storage proteins while protecting the fruit until seed maturation (Koiwa et al., 1997).

Phytopathogenic strains of bacteria belonging to several genera, including Pseudomonas, Erwinia, Xanthomonas and Ralstonia, secrete cysteine proteases during plant infection (Hotson and Mudgett, 2004). The majority of these proteases are effectors that were initially identified due to their avirulence functions leading to activation of immune responses. The Pseudomonas syringae effectors AvrPphB and AvrRpt2 cleave host target proteins as a virulence mechanism; however, detection of cleaved products by RPS5 and RPS2, respectively, induces resistance responses (Axtell and Staskawicz, 2003; Mackey et al., 2003; Day et al., 2005; Ade et al., 2007; Zhang et al., 2010). During infection, Magnaporthe grisea, the causal agent of rice blast, secretes the zinc metalloprotease Avr-Pita, which is subsequently recognized by the host cognate receptor Pi-ta, leading to the hypersensitive response (Bryan et al., 2000; Orbach et al., 2000). Surprisingly, despite the overwhelming role of proteases in virulence, no host PIs that function in plant immunity through direct inhibition of pathogen-derived effector proteases have been described. However, the Arabidopsis KUNITZ TRYPSIN INHIBITOR (KTI1) modulates pathogen-elicited cell death and specifically antagonizes necrosis caused by the phytotoxin fumonisin B1, but its mechanism of function has not been determined (Li et al., 2008). Toxins cause necrosis, with hallmarks of programmed cell death, a process that is regulated through an intricate balance of proteases and their respective PIs (Wolpert et al., 2002). In tomato, PIs have long been associated with defense, with abrogated PI expression correlating with increased susceptibility to insect herbivory and necrotrophic infection (Li et al., 2004; AbuQamar et al., 2008).

At present, relatively little is known about the role(s) of PIs in Arabidopsis defense. Here, we studied the function of the inhibitor encoded by the UNUSUAL SERINE PROTEASE INHIBITOR (UPI) gene in defense against necrotrophic fungi and insect herbivory. Induction of UPI is dependent on jasmonate (JA) and salicylic acid (SA) responses, but its expression is suppressed by exogenous application of the ethylene (ET) precursor aminocyclopropane-1-carboxylic acid and ET signaling. The upi mutant shows enhanced susceptibility to B. cinerea and A. brassicicola, and reduced tolerance to feeding by the generalist insect pest Trichoplusia ni. Unexpectedly, the upi mutant showed significantly delayed flowering, corresponding with altered expression of genes involved in the control of flowering time. Recombinant UPI inhibited proteolytic activity of the serine protease chymotrypsin, and, to a lesser degree, the cysteine protease papain. Together, our data suggest that UPI is a functional PI that positively contributes to plant development and defense.


Identification of the UNUSUAL SERINE PROTEASE INHIBITOR gene and pathways regulating its expression

The UPI gene (AT5G43580) was selected for functional analysis due to its induction in B. cinerea-infected plants in expression profiling experiments. Quantitative RT-PCR confirmed UPI induction, and its expression was further studied in defense and/or hormone response mutants (Figure 1a,b). Compared to infected wild-type, B. cinerea-induced UPI expression is higher in the npr1 mutant, which is impaired in SA responses (Cao et al., 1994), and the bik1 mutant, which has increased SA accumulation (Veronese et al., 2006), whereas lower expression was observed in the SA-deficient genotypes nahG and sid2 (Delaney et al., 1994; Wildermuth et al., 2001). These data suggest that endogenous SA is required for full UPI expression during pathogen infection. In addition, UPI induction is wholly impaired in the coi1 mutant, but is nearly four times higher in the ein2 mutant relative to infected wild-type. The ein2 mutant is impaired in ET signaling and exhibits extreme susceptibility to B. cinerea (Alonso et al., 1999; Thomma et al., 1999). COI1 is the receptor for JA and is a crucial component of plant responses to wounding, herbivory and necrotrophic pathogens (Ballare, 2011).

Figure 1.

 Regulation of UPI expression by plant hormone pathways and development.
(a, b) B. cinerea-induced UPI expression in (a) SA, JA and ET signaling mutants/transgenic lines and (b) ABA/ET response mutants/transgenic lines.
(c–e) UPI expression in response to exogenous hormone application (c), after bolting (d) and in various tissues (e). Statistical significance was determined using analysis of variance and Tukey’s test. Mean values followed by different letters are significantly different from each other (= 0.05).
Experiments were performed as described in Experimental procedures and repeated at least three times with similar results. Wt, wild-type; dap, days after planting; hpi/hpt, hours after inoculation/treatment.

Expression of UPI was also examined in 35S:ERF1 and etr1 plants to assess its repression by ET (Solano et al., 1998) (Figure 1b). ERF1 encodes a transcription factor that is directly regulated by JA and ET response pathways (Berrocal-Lobo et al., 2002). Expression was unchanged in 35S:ERF1 plants, but was significantly higher in the ET receptor mutant etr1, confirming the suppressive effect of ET on UPI induction. In healthy leaves, UPI has low basal expression that is significantly up-regulated in response to SA, MeJA and abscisic acid (ABA), but not aminocyclopropane 1-carboxylic acid, consistent with expression data from mutant plants (Figure 1c). Generally, ABA acts as a suppressor of plant defense against necrotrophic fungi, functioning in direct opposition to ET responses (Mauch-Mani and Mauch, 2005). UPI expression is almost sevenfold higher in the abi1 mutant but is marginally repressed in the abi5 mutant following infection (Figure 1b). ABI1 is involved in the negative regulation of ABA signaling (Gosti et al., 1999), whereas ABI5 positively regulates ABA responses, with loss of function resulting in ABA insensitivity (Finkelstein and Lynch, 2000). In summary, ABA and SA positively modulate UPI expression, which is dependent on COI1-mediated JA responses and negatively regulated by the ET pathway.

Finally, tissue-specific expression was studied to identify the developmental signals that control UPI expression. In leaf tissue, expression positively correlates with increased age and is highest in the first and second leaf whorls, suggesting that UPI is induced during senescence (Figure 1d,e). UPI expression is not detectable in the stem, flowers or cauline leaves, but it is marginally expressed in the roots (Figure 1e). Seeds had the highest levels of UPI expression compared with all other tissues examined.

UPI is required for resistance to necrotrophic fungi but not bacterial infection

To define the biological functions of UPI, a null T-DNA insertion mutant (upi; SAIL_30_A01) was obtained from the Arabidopsis Biological Resource Center (Figure 2a) (Alonso et al., 2003). In parallel, the mutant was transformed with genomic UPI and its upstream regulatory region (UPIpr:UPI;upi) or with UPI cDNA under the control of a constitutive promoter (35S:UPI;upi), and selected transgenic lines, identified based on UPI expression, were used for the studies described (Figure 2b). Loss of upi function enhances susceptibility to B. cinerea, but its constitutive expression (35S:UPI;upi) results in increased resistance relative to wild-type (Figure 2c–f). UPIpr:UPI;upi plants showed wild-type levels of resistance to B. cinerea, directly linking the susceptibility of the mutant to loss of UPI function. Trypan blue staining of infected tissue revealed extensive fungal growth in the mutant, but this was limited in 35S:UPI;upi plants (Figure 2e). Similarly, fungal growth, assessed based on quantitative PCR amplification of Actin A DNA of B. cinerea, was significantly enhanced in the mutant but restricted in the over-expression line (Figure 2f).

Figure 2.

UPI is required for resistance to B. cinerea.
(a) Location of upi T-DNA insertion resulting in loss of UPI function. The arrows indicate locations of primers used to assay expression.
(b) UPI expression in transgenic lines.
(c, d) Disease phenotypes after drop inoculation (c) and spray inoculation (d) with B. cinerea. Images were taken 3 days post-inoculation.
(e) Trypan blue staining of fungal hyphae in spray-inoculated leaves 48 h post-inoculation.
(f) Fungal growth 5 days after spray inoculation with B. cinerea. Quantitative PCR amplification of B. cinerea ActinA relative to Arabidopsis Actin2 genes.
Experiments were performed as described in Experimental procedures and repeated at least three times with similar results. Wt, wild-type.

UPI also contributes to resistance to the necrotrophic fungus A. brassicicola. Inoculated upi leaves developed lesions twice the size of those on inoculated wild-type, producing four times as many spores per lesion (Figure 3a–c). Resistance was restored to wild-type levels in both 35S:UPI;upi and UPIpr:UPI;upi plants, with no discernable increase in resistance observed for 35S:UPI;upi compared with UPIpr:UPI;upi (Figure 3a). The mutant was also tested for responses to the bacterial pathogen P. syringae pv. tomato (Pst) to determine the specificity of UPI function in disease resistance. upi showed no altered bacterial growth or disease symptoms in response to virulent Pst DC3000 (Figure 3d) or the non-host strain P. syringae pv. phaseolicola (data not shown). In addition, upi and 35S:UPI;upi plants were assayed for responses to Pst DC3000 avrRpm1 to determine whether UPI has a function in modulating plant programmed cell death responses. Neither loss of UPI expression nor constitutive UPI expression affected the level of bacterial growth in inoculated plants (Figure 3e). Trypan blue staining of leaves 6 h post-inoculation with Pst DC3000 avrRpm1 also revealed no obvious differences in the degree of cell death in upi and 35S:UPI;upi plants compared to wild-type (Figure 3f). Thus, the defense role of UPI appears to be restricted to necrotrophic fungal infection.

Figure 3.

 The upi mutant shows increased susceptibility to A. brassicicola but normal responses to P. syringae.
(a–c) Disease symptoms (a), mean lesion size (b) and number of spores per lesion (c) after A. brassicicola inoculation.
(d, e) Bacterial growth after inoculation with virulent P. syringae (Pst DC3000) (d) or avirulent P. syringae (Pst DC3000 avrRpm1) (e).
(f) Trypan blue staining of cell death in leaf tissue without inoculation (top row) and 6 h post-infiltration with magnesium sulfate (MgSO4, middle row) or inoculation with Pst DC3000 avrRpm1 (bottom row). The staining in MgSO4-treated leaves indicates cell death inflicted by syringe infiltration and that in Pst-inoculated leaves indicates local hypersensitive response-induced cell death.
Data in (b) and (c) represent means ± SE from a minimum of 30 disease lesions. Experiments were performed as described in Experimental procedures and repeated at least three times with similar results. Statistical analysis was performed as described in the legend to Figure 1. Wt, wild-type; dpi, days post-inoculation; CFU, colony-forming units.

The upi mutant does not show altered wound-induced resistance to B. cinerea

Wounding Arabidopsis leaves prior to inoculation induces strong but transient immunity to B. cinerea, independent of SA- and JA/ET-mediated responses (Chassot et al., 2008). Wounded upi, UPIpr:UPI;upi and 35S:UPI;upi lines displayed enhanced immunity comparable to wounded wild-type plants, suggesting that UPI has no role in wound-induced resistance to B. cinerea (Figure S1). Further, tissue taken at the site of wounding or pathogen inoculation showed increased UPI expression, but no discernable systemic induction occurred in treated plants (data not shown).

Defense gene expression and analysis of co-expressed genes

In an attempt to link UPI function in defense to specific pathways, we assessed altered B. cinerea-induced expression of PR-1 and PDF1.2, molecular markers for SA- and JA/ET-dependent defense responses in Arabidopsis, respectively (Figure S2a,b). Induction of these genes is unaltered in the upi mutant, indicating that loss of UPI does not influence the pathways leading to their expression. However, induction of the defense response genes PR-3 and PR-4 was significantly higher and abolished, respectively, in the upi mutant (Figure S2c,d). By contrast, 35S:UPI;upi plants showed significantly higher basal PR-4 expression but display wild-type induction levels following infection (Figure S2d).

Co-expression analysis ( identified KTI1 (At1g73260), a β-glucosidase gene (At1g61820) and PLEIOTROPIC DRUG RESISTANCE9 (PDR9, At3g53480) as having relatively high (>0.7) correlations of expression with UPI (Figure S3). KTI1 encodes a PI that is involved in modulating programmed cell death during disease (Li et al., 2008), and thus was considered a potential interacting partner of UPI. However, the two proteins did not interact in yeast two-hybrid (Y2H) assays (data not shown).

UPI expression is COI1-dependent and negatively-regulated by MYC2

Methyl viologen, P. syringae, wounding and B. cinerea significantly induce UPI expression in wild-type plants; however, the coi1 mutation eliminated induction regardless of treatment (Figures 1 and 4a,b). To assess UPI function downstream of COI1, we assayed its expression in the splayed (syd) and myc2 mutants. SYD encodes a JA-regulated chromatin remodeling ATPase that, similar to UPI, is required for floral transition and resistance to B. cinerea but not defense against P. syringae (Wagner and Meyerowitz, 2002; Walley et al., 2008). In response to B. cinerea, UPI expression was significantly lower in the syd mutant than in wild-type (Figure 4c). MYC2 functions as a positive regulator of JA signaling downstream of COI1 in response to wounding, herbivory and oxidative stress, but suppresses JA-mediated PDF1.2 expression during defense (Dombrecht et al., 2007; Kazan and Manners, 2008). The myc2 mutant shows increased resistance to B. cinerea and P. cucumerina (Boter et al., 2004; Lorenzo et al., 2004). Consistent with the role of MYC2 as a negative regulator of defense downstream of COI1, UPI expression is significantly higher in the myc2 mutant following B. cinerea or P. syringae inoculation (Figure 4c,d). Consistent with the increased UPI expression in inoculated myc2, VSP1, a wound-responsive gene that is positively regulated by MYC2, was up-regulated in the upi mutant (Figure 4e) (Boter et al., 2004; Lorenzo et al., 2004). Analysis of the 5′ sequence upstream of the UPI coding sequence for cis-acting regulatory elements (Higo et al., 1999) revealed three putative MYC2-binding sites (a CANNTG consensus sequence located 306, 401 and 666 nucleotides upstream of the UPI start codon), further suggesting these proteins functionally interact. The UPI promoter sequence also contains a T/G box, a regulatory element of the tomato protease inhibitor II (pin2) gene that is recognized by JAMYC2 and JAMYC10 for JA-induced pin2 expression (Boter et al., 2004). Intriguingly, basal expression of EPITHIOSPECIFYING SENESCENCE REGULATOR (ESP/ESR) is low but significantly induced following infection with B. cinerea in the upi mutant, which is in direct contrast to its pattern of expression in wild-type plants (Figure 4f). ESP/ESR is a JA-inducible protein that mediates negative crosstalk between plant responses to senescence and pathogen infection (Miao and Zentgraf, 2007). Although no differences were observed between UPIpr:UPI;upi and wild-type plants, 35S:UPI;upi plants showed significantly lower basal and B. cinerea-induced ESP/ESR and VSP1 expression, respectively (Figure 4e,f). The effects of constitutive UPI expression on ESP/ESR and VSP1 transcript levels are in contrast to those resulting from its loss of function, further supporting the effect of UPI on regulation of these genes.

Figure 4.

 Differential regulation of UPI expression by COI1 and MYC2.
(a, b) UPI expression in wild-type and coi1 plants after wounding or methyl viologen (MV) treatment (a), and after P. syringae inoculation (b).
(c, d) UPI expression in myc2 and syd mutants (c) in response to B. cinerea, and in myc2 following P. syringae inoculation (d).
(e, f) B. cinerea-induced expression of VSP1 (e) and ESP/ESR (f) in the UPI genotypes.
Experiments were performed as described in Experimental procedures and repeated at least three times with similar results. Statistical analysis was performed as described in the legend to Figure 1. Wt, wild-type; hpt/hpi, hours post-treatment/inoculation.

UPI is required for resistance to Trichoplusia ni feeding

Previous reports have established that plant defenses against necrotrophs and insect pests have functional overlap, with both responses dependent on JA and ET (AbuQamar et al., 2008). Thus, we assayed UPI function in resistance to insect herbivory using feeding trials with larvae of Trichoplusia ni, a generalist herbivore. After 10 days of feeding, upi mutant plants were almost completely defoliated, whereas 35S:UPI;upi plants showed greater tolerance to feeding compared to wild-type (Figure 5a). UPIpr:UPI;upi plants also showed enhanced resistance, but this increase was moderate relative to the wild type. Larvae feeding on upi mutants had significantly higher weights than larvae on wild-type plants (Figure 5b,c). Larvae feeding on 35S:UPI;upi plants had the lowest weights, and those feeding on UPIpr:UPI;upi plants also showed a significant decrease in weight compared to larvae fed on wild-type plants. These data provide clear genetic evidence for the role of UPI in plant defense against insect herbivory.

Figure 5.

 Constitutive UPI expression increases plant tolerance to T. ni larval feeding.
(a) Herbivory damage, (b) larval growth and (c) mean larval weight after a 10-day feeding trial on the UPI genotypes and wild-type plants. Data in (c) represent means ± SE from 30 to 45 larvae. Experiments were performed as described in Experimental procedures and repeated at least three times with similar results. Statistical analysis was performed as described in the legend to Figure 1. Wt, wild-type.

UPI has a distinct sequence relative to other PIs

The UPI cDNAs S62951 (GenBank accession number BT010789) and U62959 (Genbank accession number BT015378) were identified from the SALK collection (Yamada et al., 2003) and verified via RT-PCR using primers in the coding sequence, 5′ UTR and genomic regions (Figure S4). These analyses confirmed that the cDNA (222 bp) contains a predicted ORF (73 amino acid) corresponding to the coding sequence of UPI. UPI encodes an 8.8 kDa serine PI with features of potato inhibitor type I (PTI-I) proteins that are predicted to be members of the pathogenesis-related protein 6 (PR-6) family ( (Sels et al., 2008). However, BLAST searches against the National Center for Biotechnology Information ( and Arabidopsis databases ( using the UPI sequence returned no proteins with significant similarity. PTI-I inhibitors are proteins of approximately 8 kDa generally lacking disulfide bonds that contain a family signature sequence [FYW]-P-[EQH]-[LIV]2-G-x2-[STAGV]-x2 that is located in the N-terminus of the peptide (Habib and Fazili, 2007; Hulo et al., 2008). However, UPI lacks this family signature (Figure 6a), and protein fold-recognition software (Kelley et al., 2000) predicted that it has only 15% structural homology to a putative serine PI. Promoter comparison between UPI and other predicted Arabidopsis PR-6 genes showed no shared defense-related cis-acting regulatory elements (Lescot et al., 2002). Further, phylogenetic analyses confirmed UPI divergence from other putative PR-6 proteins (Figure 6b), the main characteristics of which are summarized in Figure 6(c).

Figure 6.

 Comparison of UPI with members of the Arabidopsis PR-6 protein family.
(a) Sequence comparison between UPI and putative Arabidopsis PR-6 proteins. Black shading indicates conserved residues, grey shading indicates residues identical to UPI; the PTI-1 signature sequence is indicated by black line. Sequences were aligned using CLUSTAL W (Thompson et al., 1994) with default gap penalties.
(b) Phylogenetic relationships between UPI and putative PR-6 proteins. Mean character distances were used to construct the unrooted neighbor-joining phylogeny (Saitou and Nei, 1987) using Phylip version 3.67 with 1000 bootstrap re-samples (Felsenstein, 2007).
(c) Summary of the main features of the Arabidopsis PR-6 family.

Recombinant UPI inhibits serine and cysteine proteases independently of a conserved aspartic acid residue in the putative reactive bond

The function of most plant PIs is based on indirect evidence, with no biochemical and genetic data showing their function. Generally, serine PIs block the hydrolytic activity of serine and/or cysteine proteases by binding to the active site of the target proteases (Bode and Huber, 1992). Inhibition is dependent on tertiary structure and is mediated by a loop containing a reactive site that physically interacts with the protease (Bode and Huber, 1992). Although it lacks significant homology with characteristic PTI-1 proteins, reverse zymography results using purified recombinant hemagglutinin-tagged UPI (Figure 7a) showed that UPI is a functional PI that is able to inhibit the proteolysis of an artificial substrate (gelatin) by chymotrypsin, and, to a lesser degree, by papain (Figure 7b). Chymotrypsin and papain are serine and cysteine proteases, respectively, that are routinely used for in vitro inhibition assays. UPI PI activity is visible as a blue Coomassie-stained band of undigested gelatin that is not visible in gels lacking UPI or positive control PIs. Control reactions containing proteases and their corresponding inhibitors (chymostatin for chymotrypsin, and antipain for papain) show similar inhibitory activity.

Figure 7.

 Reverse zymography revealing UPI inhibition of chymotrypsin and papain proteolytic activity.
(a, b, d) Coomassie-stained SDS–PAGE gels showing purified recombinant UPI–HIS with anti-HIS immunoblotting (a), UPI–HIS inhibition of gelatin proteolysis by chymotrypsin and papain (b), and inhibitory activity of the substitution mutant UPID45A compared with UPI (d).
(c) UPI sequence indicating the position of the UPID45A amino acid substitution (upper) and the sequence of a typical PTI-1 protein (lower).
(e) Immunoblot showing UPI accumulation in leaf tissue in response to methyl jasmonate (MeJA).
(f) Immunoblot of intracellular (left column) and extracellular (right column) UPI accumulation following water or MeJA treatment.
Experiments were performed as described in Experimental procedures and repeated at least three times with similar results. In (b) and (d), chymostatin and antipain were used as positive controls for enzymatic inhibition of the serine protease chymotrypsin and the cysteine protease papain, respectively. In (f), apoplastic fluid was isolated from leaves which were then dried under vacuum for subsequent use in determining intracellular protein levels. Immunoblot detection of the cytoplasmically localized MAP kinase MPK4 from the same samples used for UPI analysis served as an internal control. Ponceau S staining was used as a loading control.

PIs function in a conformation-dependent manner, with proper folding allowing exposure of active residues that bind to and block the hydrolytic activity of target proteases (Bode and Huber, 1992; Habib and Fazili, 2007). Despite its limited homology to PIs, UPI contains a conserved aspartic acid residue in the C-terminus (Figure 7c) that was found to be part of a reactive bond contributing to inhibitor structure in eglin c, a PI that is highly homologous to the PTI-I family (Svendsen et al., 1982; Bode et al., 1986; McPhalen and James, 1988). We substituted the aspartic acid residue of the reactive bond with alanine to further define UPI function relative to other PIs (Figure 7c). Reverse zymography revealed no altered inhibition activity for the mutated UPI protein (UPID45A) compared to UPI (Figure 7d).

JA induces inter- and intracellular UPI accumulation

Various prediction programs have returned contrasting results regarding the localization of UPI. We transiently expressed UPI–GFP in Nicotiana benthamiana to determine its subcellular localization. In comparison to the GFP control, UPI–GFP had no clear or specific localization, which may be attributed to the size discrepancy between UPI and GFP (Figure S5). However, immunoblot analyses of transgenic upi plants expressing hemagglutinin-tagged UPI revealed that MeJA induces both intracellular and extracellular accumulation of UPI (Figure 7e,f). Expression of hemagglutinin-tagged UPI in the upi mutant complemented phenotypes of upi, showing that the epitope tag does not affect biological function (Figure S6).

UPI does not directly inhibit fungal growth in vitro

Numerous plant PIs have been shown to directly inhibit fungal growth (Joshi et al., 1999; Pernas et al., 1999; Martinez et al., 2005). Due to its small size, extracellular accumulation and the increased susceptibility of the mutant to pathogens and insect pests, we hypothesized that UPI may be involved in direct inhibition of exogenous targets. However, no apparent differences in fungal growth were observed on V8, water agar or potato dextrose agar fungal growth media supplemented with recombinant UPI–GST compared with unsupplemented plates or media supplemented with GST (glutathione S-transferase) alone (Figure S7). Additionally, sterile filter papers (1 cm2) saturated with 5 μg of UPI–GST or GST placed on fungal growth media containing approximately 500 conidia ml−1 of B. cinerea or A. brassicicola revealed no altered zones of growth inhibition relative to buffer-treated controls (plates were visually assessed every 2 days for 14 days; data not shown). Comparisons of conidial germination, hyphal growth and general fungal morphology also revealed no differences between B. cinerea or A. brassicicola conidia incubated with purified UPI–GST or GST (data not shown).

UPI regulates flowering time independently of photoperiod

The upi mutant exhibits significantly delayed flowering, independently of photoperiod (Figure 8). Under 12 h light conditions, the upi mutant starts to bolt 10 days later than wild-type, producing significantly more rosette leaves (Figure 8b). Consistently, the upi mutants shows lower expression of SOC1, a downstream promoter of flowering, and elevated levels of MAF1, a floral repressor and FLC homolog (Figure 8c,d) (Ratcliffe et al., 2001; Hepworth et al., 2002). UPIpr:UPI;upi and 35S:UPI;upi plants flowered at a similar time to wild-type, and showed wild-type levels of SOC1 and MAF1 expression (Figure 8), indicating that loss of UPI function is the sole cause of delayed flowering in the mutant. Interestingly, many Arabidopsis mutants with significantly reduced UPI expression (Zimmermann et al., 2004) show altered floral transition and development (Table S1). UPI expression is also significantly affected by light conditions, photoperiod and temperature shifts, all major environmental factors that affect flowering (Imaizumi and Kay, 2006). Due to its small size (8.8 kDa) and extracellular accumulation, we hypothesize that UPI may function as a transmissible signal in the regulation of flowering. However, reciprocal grafts between wild-type and upi plants revealed no differences in time of flowering between grafted plants and non-grafted controls (Figure S8).

Figure 8.

 The upi mutant shows delayed flowering associated with altered MAF1 and SOC1 expression.
(a, b) Delayed flowering phenotype of the upi mutant (a), and number of rosette leaves at bolting under various photoperiods (b).
(c, d) Expression of SOC1 (c) and MAF1 (d) in the UPI genotypes.
Experiments were performed as described in Experimental procedures and repeated at least three times with similar results. Statistical analysis was performed as described in the legend to Figure 1. Wt, wild-type; hdl, hours of day light.

The upi mutation does not impair oxidative stress tolerance or hormone-mediated growth responses

Based on its induction by SA, JA and ABA, as well as cell death and defense-inducing compounds, we explored possible connections between UPI function and hormones that mediate Arabidopsis biotic/abiotic stress responses. The upi mutation does not affect germination or growth on media supplemented with aminocyclopropane 1-carboxylic acid, SA, MeJA, gibberellic acid or ABA (Figures S9 and S10) or alter responses to oxidative stress caused by salinity or methyl viologen (Figures S10 and S11). Methyl viologen and necrotrophic infection both induce formation of reactive oxygen species in plants, suggesting that the susceptibility of the upi mutant to fungal infection is not due to decreased oxidative stress tolerance (Govrin and Levine, 2000; Suntres, 2002; Mengiste et al., 2003). This is further supported by its wild-type growth responses on hydrogen peroxide-supplemented medium (Figure S10).


We provide genetic and biochemical data that establish that Arabidopsis UNUSUAL SERINE PROTEASE INHIBITOR (UPI) is a component of defense against necrotrophic pathogens and insect pests that also influences flowering time. Loss of upi function results in enhanced susceptibility to necrotrophic fungi and feeding by Trichoplusia ni, with ectopic expression increasing tolerance to both. Consistent with its function, UPI expression is modulated by exogenous application of defense-mediating hormones, showing induction in response to MeJA, SA and ABA treatment. Induction of UPI is dependent on functional JA signaling but is suppressed by ET responses, two Arabidopsis pathways with typically synergistic interactions. Delayed flowering in the mutant is accompanied by altered SOC1 and MAF1 expression suggesting that UPI has a dual function in defense and regulation of flowering time.

UPI was recently classified as a PR-6 protein (Sels et al., 2008). The PR-6 family constitutes a sub-class of serine PIs with characteristics of potato/tomato type I PIs (Sels et al., 2008). However, sequence analyses reveal limited similarity between UPI and characteristic potato/tomato type I PIs, other PR-6 proteins, and all other Arabidopsis proteins. Despite this lack of similarity, UPI is a bi-functional PI that is able to inhibit the serine protease chymotrypsin as well as weakly block activity of the cysteine protease papain. As PI activity is dependent on peptide sequence and tertiary conformation, its sequence divergence, coupled with the phenotypes resulting from its loss of function, suggest that UPI does not share functional redundancy with other PIs (Bode and Huber, 1992). However, constitutive expression of At2g38870, which encodes another putative PR-6 family member, also confers enhanced resistance to B. cinerea, but its potential function in Arabidopsis defense responses has not been examined (Chassot et al., 2007). The JA-induced extracellular accumulation of UPI also suggests that it may serve as an intercellular signal or directly target pathogen/insect-derived proteases.

UPI expression is dependent on the JA receptor COI1, which is required for defense against necrotrophs, herbivory and wounding – all factors that elicit UPI expression (Ballare, 2011). Constitutive expression of ERF1, a downstream target of COI1 and a point of convergence between JA and ET signaling, does not affect B. cinerea-induced UPI expression (Ferrari et al., 2003). However, the ET-insensitive mutants ein2 and etr1 (Chang et al., 1993; Alonso et al., 1999) have significantly higher UPI induction, suggesting that ET acts as a negative regulator of UPI expression. JA and ET typically function synergistically in defense; however, they also regulate independent plant responses, with negative interactions similar to their regulatory impact on UPI (Glazebrook, 2005; Lorenzo and Solano, 2005). ET antagonizes local wound-induced JA responses by repressing a subset of JA-regulated genes (Rojo et al., 1999). Induced expression of this subset of genes was found to be significantly higher in ein2 and etr1 mutants (Rojo et al., 1999). MYC2 functions downstream of COI1 in the differential regulation of JA signaling, acting as a positive regulator of wound, herbivory and oxidative stress responses that directly suppresses JA/ET-mediated defense to necrotrophic infection (Lorenzo et al., 2004; Dombrecht et al., 2007; Kazan and Manners, 2008). MYC2 also suppresses both P. syringae- and B. cinerea-induced UPI expression. Interestingly, similar to UPI, mutation in MYC2 results in delayed flowering (Yadav et al., 2005). SYD, which is required for floral transition and defense against necrotrophs, is also required for full UPI expression in response to B. cinerea (Wagner and Meyerowitz, 2002; Walley et al., 2008). SYD binds to the promoter of MYC2, with loss of function resulting in decreased levels of MYC2 and altered expression of JA/ET-responsive genes (Walley et al., 2008). In summary, UPI is controlled by an unusual regulatory loop that involves negative interactions of ET and JA downstream of COI1, as well as SA- and ABA-mediated responses.

The upi mutant shows higher induction of VSP1, a defense gene that is positively regulated by MYC2, following infection (Dombrecht et al., 2007). Generally, VSPs have been regarded as nutrient reserves that are cleaved by proteases into amino acids for plant assimilation during growth (Staswick, 1994). PIs regulate this nutrient mobilization by blocking protease activity, and therefore are generally abundant in storage tissues with a low presence in leaf tissue (Green and Ryan, 1972; Muntz et al., 2001). UPI follows typical PI expression patterns, exhibiting increased accumulation in seed and senescent tissues. The presence of PIs in storage tissues has also been suggested to serve as a defense mechanism against pathogen infection and herbivory until maturity, consistent with UPI function (Brown and Adikaram, 1983; Koiwa et al., 1997). Recently, COI1 was shown to regulate VSP1 accumulation in flower tissue (Chua et al., 2010). Potentially, UPI may function in COI1-dependent VSP1 regulation by inhibiting proteases that target it for degradation. Consistent with this idea, VSP1 expression decreases with plant maturity whereas that of UPI increases with age, suggesting an inverse functional relationship (Matthes et al., 2008). Basal and B. cinerea-induced expression of ESP/ESR is also affected by upi mutation. ESP/ESR expression is highly induced from very low levels in the mutant following infection, in direct opposition to its wild-type pattern of expression. Intriguingly, ESP/ESR expression is induced by JA and is dependent on JASMONATE RESISTANT1 (JAR1), but shows constitutively high expression in the coi1 mutant (Miao and Zentgraf, 2007). Thus, like UPI, ESP/ESR appears to be differentially regulated by different branches of the JA pathway. Together with WRKY53, ESP/ESR regulates negative crosstalk between senescence and responses to pathogen infection (Miao and Zentgraf, 2007). Potentially, based on the increased induction of both VSP1 and ESP/ESR in the mutant, loss of UPI function may result in over-activation of JA signaling. As JA signaling related to wound responses directly antagonizes ET-dependent defenses, the increased susceptibility of the mutant may be a result of altered ET signaling. This idea is further supported by the abolished expression of PR-4, a marker of the ET signaling pathway, in upi plants. Thus, future investigation of the effects of UPI on ESP/ESR expression and other JA/ET-regulated genes may clarify UPI function in plant defense, potentially shedding light on the regulation of negative interactions between hormone signaling pathways and senescence programming.

Similar to JA, ABA has a positive effect on UPI expression and has previously been shown to induce PI genes (Hildmann et al., 1992; Carrera and Prat, 1998). ABA mediates defense responses, with both positive and negative effects on resistance, but is also a major regulator of abiotic stress responses (Fujita et al., 2006; Ton et al., 2009). During wound responses, ABA activates MYC2 expression in a COI1-dependent manner and induces JA accumulation (Lorenzo et al., 2004). ABA and ET have mutually antagonistic functions in response to pathogens, seed germination, growth and development (Anderson et al., 2004; Cheng et al., 2009). ABI1 is a negative regulator of ABA responses, with loss of function resulting in phenotypes mirroring those of etr1, including susceptibility to necrotrophs (Murata et al., 2001; Ferrari et al., 2003; Guimaraes and Stotz, 2004; Desikan et al., 2005; Korolev et al., 2008). This is consistent with the altered expression of UPI in both backgrounds and its negative regulation by ET. Additionally, mutation of ABI5, which encodes a positive regulator of ABA responses, decreases B. cinerea-induced UPI expression, further indicating that ABA contributes positively to UPI regulation (Brocard et al., 2002).

Based on its in vitro PI activity and induction by elicitors of cell death, UPI may function in the containment of necrosis during disease. Cysteine proteases mediate host cell death, including necrosis and the hypersensitive response (Solomon et al., 1999; Piszczek and Gutman, 2007). Initiation of the hypersensitive response in soybean (Glycine max) and oat (Avena sativa) was shown to require cysteine proteases and two serine proteases exhibiting caspase-like activity, respectively (Solomon et al., 1999; Alvarez, 2000; Coffeen and Wolpert, 2004). Treatment of Avena sativa with PIs was sufficient to block cell death induced by victorin, a toxin produced by the necrotrophic fungus Cochliobolus victoriae (Coffeen and Wolpert, 2004). The hypersensitive response and oxidative stress-induced cell death were abolished in soybean cells ectopically expressing cysteine protease inhibitor genes (Solomon et al., 1999). B. cinerea causes necrosis with features similar to programmed cell death and accumulation of reactive oxygen species, which serve as signals for the hypersensitive response-like cell death that occurs during infection (Govrin and Levine, 2000). UPI induction in response to pathogens and exogenous SA is consistent with its role as a suppressor of B. cinerea-induced necrosis. However, the wild-type susceptibility and cell death responses to Pst DC3000 avrRpm1 of plants constitutively expressing UPI and the upi mutant suggest that UPI function in defense is not related to the regulation of cell death. This is further supported by the normal oxidative stress tolerance of upi plants. UPI may target an endogenous protease that promotes disease-associated cell death. However, based on its JA-induced secretion and apparent lack of involvement in hypersensitive response-related responses, it appears more likely that UPI may target pathogen proteases that promote disease, although we were unable to demonstrate in vitro fungal growth as the target proteases may only be expressed during pathogenesis.

In addition to defense, UPI regulates floral transition as evidenced by the significantly delayed flowering of the mutant. Intriguingly, over-expression of KUNITZ-TYPE TRYPSIN INHIBITOR1 (KTI1), the only other studied PI gene involved in defense against necrotrophs, results in early flowering (Li et al., 2008; Kim et al., 2009). upi plants show reduced expression of SOC1, which encodes a downstream promoter of flowering (Hepworth et al., 2002). The soc1 mutant exhibits late flowering independently of photoperiod, thus the delayed flowering of the upi mutant is likely to result from decreased SOC1 expression (Borner et al., 2000). Also, the upi mutant shows elevated levels of MAF1, a known floral repressor and FLC homolog (Ratcliffe et al., 2001). Based on public microarray data (Zimmermann et al., 2004), the transgenic Arabidopsis lines atx1, lfy, 35S::amiR-lfy-1, agl104, elf8 and se, which are altered with respect to floral development and/or transition, also show significantly repressed UPI expression.

As UPI is a small secreted protein, we hypothesize that it may serve as a mobile signal to regulate flowering or inhibit fungal proteases. However, grafting and in vitro anti-fungal assays did not indicate such functions. This suggests that UPI may require endogenous co-factors or lose specificity when produced as a recombinant protein, as inclusion of an epitope tag may significantly alter UPI structure and thus function. Additionally, artificially grown cultures have vastly different expression profiles compared to fungal growth in planta, with the degree of disease progress also affecting expression. Thus, the function of UPI as an anti-fungal agent may be contingent on specific recognition events occurring at the host–pathogen interface. The lack of in vitro inhibition may also indicate that UPI has endogenous targets and only an indirect function in defense.

Overall, the role of UPI in the regulation of defense against necrotrophic infection, insect herbivory and flowering time is clear; however, its mechanism and relationship to COI1-mediated JA responses require further investigation. The proposed model of UPI regulation and function in defense is summarized in Figure 9. COI1-dependent JA responses strictly regulate UPI induction, whereas its expression is suppressed by ET and MYC2 regulated responses. ABA and SA also function in the positive regulation of UPI expression. Whereas PDF1.2 is an indicator of their synergistic function, UPI appears to be an indicator of negative interaction between the ET and JA pathways. Overall, UPI has the potential to serve as an endogenous intercellular or extracellular signal contributing to plant development and defense through regulation of endogenous and/or pathogen-derived targets. Identification of these targets in the future will help in elucidating the functions of UPI.

Figure 9.

 Proposed model showing pathways regulating UPI function in Arabidopsis defense against insects and pathogens.

Experimental procedures

Plant maintenance

Plants were grown in soil under fluorescent light (150 μE m−2 sec−1) at 23 ± 2°C with 60% relative humidity and a 12 h light/12 h dark cycle. For flowering experiments, short-day (8 h light/16 h dark) or long-day (16 h light/8 h dark) growth conditions were used. Transgenic selection, germination and growth assays were performed as described previously (Dhawan et al., 2009). Grafting experiments were performed as previously described without addition of benomyl (methyl 1-(butylcarbamoyl)-2-benzimidazolecarbamate) and 6-benzylaminopurine (Rus et al., 2006). Uncut and self-grafted wild-type and upi plants were used as controls.

Feeding trials

Trichoplusia ni eggs were hatched on Bio-Serv Lepidoptera diet (, and 2 days after egg transfer to diet trays, two larvae per plant were placed on 30-day Arabidopsis diet (Bio-Serv). Trays were covered with clear lids containing air holes, larvae were left to feed for 10 days, and their weights were determined.

Fungal cultures and disease assays

Maintenance of fungal cultures and disease assays were performed as previously described (Dhawan et al., 2009). The culture and disease assays for P. syringae strains were performed as described by Mengiste et al. (2003). Wound-induced resistance to B. cinerea was assayed as described previously (Chassot et al., 2008). Trypan blue staining of inoculated plant tissue was performed as described by Ek-Ramos et al. (2010).

Generation of UPI transgenic lines and mutant identification

UPI over-expression lines were generated by cloning the full length UPI cDNA (AT5G43580) behind the CaMV 35S promoter in a pCAMBIA-derived vector. UPI complementation lines were generated by cloning the full genomic sequence of UPI and approximately 800 bp of upstream sequence into pCAMBIA1391 ( Constructs were transferred into Agrobacterium strain GV3101 and transformed into Arabidopsis (Clough and Bent, 1998). The upi mutant allele (SAIL_30_A01) was identified from a segregating population obtained from the Arabidopsis Biological Resource Center using T-DNA- and gene-specific primers (Sessions et al., 2002). The previously described abi1 (SALK_076309), abi5 (CS8105), 35S:ERF1 (CS6142) and etr1 (CS3070) transgenic lines were obtained from the Arabidopsis Biological Resource Center (Chang et al., 1993; Solano et al., 1998; Gosti et al., 1999; Finkelstein and Lynch, 2000; Berrocal-Lobo et al., 2002). Primers used are listed in Table S2.

RNA extraction, blots and expression analyses

RNA extraction, cDNA synthesis and quantitative RT-PCR were performed as previously described (Dhawan et al., 2009). Quantitative RT-PCR was performed using gene-specific primers, with Arabidopsis Actin2 as an endogenous reference for normalization. A minimum of three technical replicates were used for each sample, with a minimum of two biological replicates for quantitative RT-PCR. Expression levels were calculated by the comparative cycle threshold method (Applied Biosystems) with normalization to the control as previously described (Bluhm and Woloshuk, 2005). For analysis of induced expression, plants were treated as described previously (Veronese et al., 2006). Wounding was performed by pressing approximately 60% of the leaf surface area between serrated forceps. Primers used are listed in Table S2.

Recombinant protein purification and reverse zymography

The UPI cDNA or a mutant derivative was cloned into pGEX 4T-1 (GE Healthcare, and pRSETA (Invitrogen, Expression and purification of the fusion proteins was performed as described by the plasmid manufacturers. The specificity of purification was assayed by anti-histidine immunoblot (GE Healthcare), and PI activity assayed by reverse zymography (Le and Katunuma, 2004). α-chymotrypsin and papain (Sigma-Aldrich, were used as representative serine and cysteine proteases, respectively, and antipain and chymostatin were used as positive controls of protease inhibition. Primers used are listed in Table S2.

Protein isolation and analysis

Hemagglutinin-tagged UPI transgenic plants were sprayed with 1 mM MeJA or water for 3 h. Treated leaves were subsequently infiltrated with protein extraction buffer (50 mm HEPES, 5 mm EDTA, 5 mm EGTA, 25 mm NaF, 50 mmβ-glycerol phosphate disodium salt pentahydrate, 2 mm DTT and 10% glycerol, pH 7.5), blotted to remove excess external liquids, and centrifuged under vacuum for 2 min (8000 g at room temperature) to isolate fluid from the apoplast. Following intercellular fluid isolation, tissues were dried under vacuum for 20 min, and ten leaf discs (3 mm diameter) from a minimum of six leaves per sample were used for total protein extraction. Samples were analyzed by immunoblotting using an anti-hemagglutinin antibody (Covance, Immunoblot detection of the cytoplasmically localized MAP kinase MPK4 was used as an internal cellular marker to validate these methods. Ponceau S total protein staining was used as a loading control.

Fungal growth inhibition assays

V8 medium (36% V8 juice,, 0.2% CaCO3, 2% Bacto-agar,, potato dextrose agar or water agar plates supplemented with 5–7 μg recombinant proteins were used for fungal growth assays. GST- and HIS-supplemented plates were used as controls. Each plate was inoculated with a 2 mm agar punch from a 10-day-old B. cinerea culture or with conidia (350 000 spores ml−1) grown on V8 media. Sterile filter paper spotted with 5 or 10 μg of the recombinant proteins was placed on media containing approximately 500 spores ml−1 to assay zones of fungal inhibition.

Yeast two-hybrid assays

In vitro interaction between UPI and At1g73260 was assessed using the GAL4 system according to the manufacturer’s instructions (Stratagene, Primers used are listed in Table S2.


This research was funded by National Science Foundation grant IOB-0749865 to T.M. We thank Dr Zhibing Lai for comments on the manuscript, Dr Synan AbuQamar (United Arab Emirate University, Department of Biology, P.O.Box 17551, Al-Ain, UAE) for help in initial isolation of the upi mutant, and Dr Scot Peck (University of Missouri, Columbia, MO) for the MPK4 antibody.