Fungal LysM effectors that comprise two LysM domains bind chitin through intermolecular dimerization

Chitin is a polymer of β-(1,4)-linked N-acetyl-D-glucosamine (GlcNAc) and a major structural component of fungal cell walls that acts as a microbe-associated molecular pattern (MAMP) that can be recognized by plant cell surface-localized pattern recognition receptors (PRRs) to activate a wide range of immune responses. In order to deregulate chitin-induced plant immunity and successfully establish their infection, many fungal pathogens secrete effector proteins with LysM domains. We previously determined that two of the three LysM domains of the LysM effector Ecp6 from the tomato leaf mould fungus Cladosporium fulvum cooperate to form a chitin-binding groove that binds chitin with ultra-high affinity, allowing to outcompete host PRRs for chitin binding. In this study, we describe functional and structural analyses aimed to investigate whether LysM effectors that contain two LysM domains bind chitin through intramolecular or intermolecular LysM dimerization. To this end, we focus on MoSlp1 from the rice blast fungus Magnaporthe oryzae, Vd2LysM from the broad host range vascular wilt fungus Verticillium dahliae, and ChElp1 and ChElp2 from the Brassicaceae anthracnose fungus Colletotrichum higginsianum. We show that these LysM effectors bind chitin through intermolecular LysM dimerization, allowing the formation of polymeric complexes that may precipitate in order to eliminate the presence of chitin oligomers at infection sites to suppress activation of chitin-induced plant immunity. In this manner, many fungal pathogens are able to subvert chitin-triggered immunity in their plant hosts.

groove that binds chitin with ultra-high affinity, allowing to outcompete host PRRs for 23 chitin binding. In this study, we describe functional and structural analyses aimed to Chitin is a homopolymer of β-(1,4)-linked N-acetyl-D-glucosamine (GlcNAc) and a major 36 structural component of fungal cell walls (Free, 2013;Lenardon et al., 2010). 37 Additionally, chitin has been characterized as a fungal microbe-associated molecular 38 pattern (MAMP) that can be recognized by plant cell surface-localized pattern 39 recognition receptors that contain extracellular LysM domains (LysM-PRRs) (Rovenich 40 et al., 2016;Sanchez-Vallet et al., 2015;Zhang et al., 2007;Zipfel, 2008). Upon defence-related genes that include those encoding hydrolytic enzymes such as chitinases 45 in order to halt fungal invasion (Altenbach and Robatzek, 2007;Boller and Felix, 2009;46 Felix et al., 1993;Jones and Dangl, 2006;Sanchez-Vallet et al., 2015). LysM-PRRs have (1.72 µM) and recruits AtLYK4 and AtCERK1 upon chitin elicitation to form a tripartite 50 receptor complex to initiate chitin signalling (Cao et al., 2014). AtCERK1 was found to 51 bind chitin directly as well, albeit with approximately 200-fold lower affinity than 52 AtLYK5 (Cao et al., 2014;Miya et al., 2007;Petutschnig et al., 2010). Moreover, a crystal 53 structure of the ectodomain of AtCERK1 revealed that only one out of its three LysMs 54 (LysM2) binds chitin (Liu et al., 2012). 55 To avoid chitin-induced immune responses, successful fungal pathogens evolved 56 various strategies to either protect fungal cell wall chitin against hydrolysis by host 57 enzymes, or prevent the activation of plant immunity by fungal cell wall-derived chitin 58 oligomers (de Jonge et al., 2010;Rovenich et al., 2014;Sanchez-Vallet et al., 2015). A 59 well-studied fungus for which several strategies to deal with chitin-triggered immunity 60 have been characterized is Cladosporium fulvum, the fungus that causes leaf mould 61 disease of tomato. C. fulvum secretes the Ecp6 effector protein during host colonization, 62 which contains three LysMs and binds chitin oligosaccharides with ultra-high affinity, to 63 prevent the activation of chitin-induced plant immune responses (Bolton et al., 2008;de 64 Jonge et al., 2010). A crystal structure of Ecp6 revealed that two of its three LysMs 65 cooperate to form a composite chitin-binding groove that binds chitin through 66 intrachain LysM dimerization (Sanchez-Vallet et al., 2013). Based on the functional analysis of C. fulvum Ecp6, it has been proposed that the 101 ability to suppress chitin-triggered immunity resides in the ability to bind chitin with Three-dimensional structure prediction of LysM effectors with two LysM domains 116 It has previously been determined that MoSlp1 from M. oryzae, Vd2LysM from V. dahliae, 117 and ChElp1 and ChElp2 from C. higginsianum contain two LysM domains, bind chitin and 118 suppress chitin-induced host immunity (Kombrink et al., 2017;Mentlak et al., 2012;119 Takahara et al., 2016). Their length varies from a minimum of 145 aa (Vd2LysM) to a 120 maximum of 176 aa (ChElp2), with the molecular weight of the mature proteins ranging 121 from 14.24 to 16.14 kDa (Fig. 1A). An amino acid sequence alignment of the LysM  The most direct method to reveal the chitin-binding mechanism of a LysM effector is by 153 determination of a three-dimensional protein structure in the presence of chitin, for 154 instance by X-ray crystallography. This strategy requires a protein crystal of sufficient 155 size and quality to be used in an X-ray diffraction experiment, which in turn requires 156 highly pure protein of a sufficiently high concentration. To this end, heterologous 157 production of each of the LysM effectors as N-terminally 6×His-FLAG-tagged fusion 158 protein was performed using Pichia pastoris as a yeast expression system. After 159 purification from the culture filtrate, the LysM effectors were subjected to protein 160 polyacrylamide gel analysis, revealing that only Vd2LysM migrated as expected based on 161 its predicted molecular weight (Fig. 1AB). Interestingly, the three other proteins however, glycans can greatly hamper crystal packing since they may prevent or reduce 167 favourable molecular contacts between protein molecules. Moreover, glycosylation may cause microheterogeneity in protein solutions that affects protein ordering as well 169 (Baker et al., 1994;Davis et al., 1993;Tang et al., 2019). On the other hand, glycosylation 170 may be explicitly required for proper protein folding and/or aid in crystal growth by 171 forming critical intermolecular contacts and thus, does not a priori hinder 172 crystallization (Mesters et al., 2007).   (Table S1).  The DLS heatmaps exhibited extremely heterogenous particle size distributions for each 216 of the LysM proteins. In particular, the particle size distribution for MoSlp1 and ChElp2 217 was quite heterogenous and ranged from 10 nm to 100 nm (Fig. 4A), which is 218 significantly larger than the expected size of 1-3 nm for a protein with a molecular 219 weight of approximately 16 kDa. Although less heterogenous, ChElp1 mostly occurred as particles of around at 100 nm, which again points towards a significant degree of 221 aggregation (Fig. 4A). Finally, Vd2LysM occurred as a heterogenous population of particles of 100 nm and larger. The heterogeneity of the four protein preparations 223 together with the relatively large particle size is likely to negatively impact crystal 224 formation (Niesen et al., 2008;Price 2nd et al., 2009). 225 In order to improve protein solubility and particle size distribution, gel filtration found to be already present in the Ecp6 and Mg1LysM crystals, suggesting that they were derived from the cell wall of yeast. In this study, four P. pastoris-produced LysM proteins 248 were directly subjected to initial screening using commercial crystallization kits PACT 249 premier™, Salt RX , Index TM , PEG RX and PEG/Ion screen (96 conditions/kit) with the 250 original concentrations (Table S1). Because we observed instant heavy precipitations in 251 more than half of the conditions, the four LysM protein preparations were diluted to half 252 the original concentrations and subjected to the initial screening again. Unfortunately, 253 none of these attempts yielded any genuine protein crystals. Subsequently, we pre-254 incubated the LysM proteins with chitinhexaose in molar ratios of 3:1 and 1:1 255 (protein:chitin) and subjected them to the initial crystallization screening again.

256
However, even after one year, none of the conditions developed genuine protein crystals.

257
To promote crystallization, active small molecules, traditionally referred to as  Finally, Vd2LysM and ChElp2 were produced in E. coli and subjected to an initial 265 screening in the absence of exogenously added chitin and after pre-treatment with 266 chitinhexaose in molar ratios of 3:1 and 1:1 (protein:chitin) using the five commercial kits, 267 and also subjected to the additive screen kit in the two different buffers. Unfortunately, also 268 these attempts were in vain.  (Fig. 6), further addition of chitin to a protein:chitin ratio of 1:1 fully shifted the 284 dominant ChElp2 particle size towards 100 nm (Fig. 6). This finding strongly suggests 285 that chitin addition mediates intermolecular LysM dimerization, leading to the 286 formation of polymeric protein complexes.

287
As a second, independent line of evidence for polymerization, we hypothesized 288 that if ChElp2 undergoes chitin-induced polymerization, we should be able to precipitate 289 polymeric complexes during centrifugation. Thus, with Ecp6 as a negative control, we 290 incubated ChElp2 overnight with chitohexaose and subsequently centrifuged the 291 samples at 20,000 g in the presence of 0.002% methylene blue to visualize the protein.

292
Indeed, a clear protein pellet appeared when ChElp2 was incubated with chitin, but not 293 in the control treatment without chitin, nor in the Ecp6 samples (Fig. 7). Next, we assessed whether a similar precipitation in the presence of chitin, as evidence for 295 polymerisation, could be obtained for MoSlp1 and Vd2LysM as well. Indeed, this appeared to be the case (Fig. 7). Collectively, these data confirm the occurrence of chitin-induced 297 polymerisation of LysM effectors that comprise two LysMs, and prove that intermolecular dimerization (Fig. 5, , 2011;Wlodawer et al., 2017). In this 316 study, we tried to address as many factors with respect to protein quality as possible, 317 but our attempts to obtain protein crystals failed nonetheless.

318
Obviously, absence of crystal formation does not prove that crystal formation is 319 impossible. However, the lack of crystal formation inspired our further thoughts about 320 LysM effector chitin binding. Theoretically, we anticipated that two possible substrate-

459
The authors declare no conflict of interest exists.  as a colour scale heat map ranging from blue (lowest abundance) to red (highest 502 abundance) for a particle size range of 1 nm to 100 µM.