Biotrophic interactions disentangled: In situ localisation of mRNAs to decipher plant and algal pathogen – host interactions at the single cell level

Plant-pathogen interactions follow spatiotemporal developmental dynamics where gene expression in pathogen and host undergo crucial changes. It is of great interest to detect, quantify and localise where and when key genes are active or inactive. Here, we adapt single molecule FISH techniques to demonstrate presence and activity of mRNAs using phytomyxids in their plant and algal host from laboratory and field materials. This allowed to monitor and quantify the expression of genes from the clubroot pathogen Plasmodiophora brassicae, several species of its Brassica hosts, and of several brown algae, including the genome model Ectocarpus siliculosus, infected with the phytomyxid Maullinia ectocarpii. We show that mRNAs are localised along a spatiotemporal gradient, thus providing proof-of-concept of the usefulness of these methods. These methods are easily adaptable to any interaction between microbes and their algal or plant host, and have the potential to increase our understanding of processes underpinning complex plant-microbe interactions.


38
Thanks to a series of technological advances over the last years, it has become clear that many 39 biological processes within and between organisms are best studied at the single cell level (e.g. about nutrients to be exchanged. Hence, gene expression changes rapidly at the cellular level, 49 and can differ between neighbouring cells (Buxbaum et al., 2015). Unless they are combined 50 with expensive or time-consuming techniques such as laser-assisted microdissection (Schuller 51 et al., 2014), widely used methods such as qPCR or RNAseq. These are not able to account for 52 spatial and temporal heterogeneity between the cells, and therefore are impracticable to study 53 single-cell changes at the scale of organs or macro-organisms. However, transformation or 54 genetic manipulation remains inaccessible for a wide range of pathogens or hosts, especially 55 those that cannot be grown in the laboratory and/or for which only very limited genetic 56 information is available (Libault et al., 2017). Furthermore, current work on plant and algal 57 microbiomes makes increasingly clear that functional studies require to take into account 58 complex microbial communities which include a huge diversity of "non-model" organisms,    (Maier et al., 2000). Its availability in 102 laboratory culture makes it a good model to start deciphering the interaction between 103 phytomyxids and their marine hosts.

104
Despite their importance as plant and algal pathogens, Phytomyxea have been difficult to study, 105 mostly because they cannot be cultured without their host and because they have a complex

138
Genes were selected on the basis of available biological background, to allow to not only test 139 and validate FISH methods, but to also validate the feasibility and usefulness of these methods 140 to disentangle biological information. The following P. brassicae genes were selected (i) the  RNase contamination than using a cryotome (Reichert-Jung, Frigocut 2800, Suppl. Note S1).

314
FISH signals in individual stack slices appeared dotted and therefore we used the "find maxima" 315 process in ImageJ to segment and count these dots. With the help of the "find stack maxima" 316 macro this process was automated for every slice in a CLSM stack (Suppl. Fig. S3, S4). CLSM PbBSMT mRNAs were detected did also appear to move from one cell to the next (Suppl. Fig.   373 S6, S7).

374
The number of signal maxima per µm³ was higher with RCA FISH (median 4,47 x 10 -02 +/-375 RSD 3,01 x 10 -02 signal maxima µm -3 , n=9) than with smFISH (median 9,38 x 10 -03 +/-RSD  Fig. S1, S2) Microscopic analyses 388 We also compared CLSM and spinning-disk microscopy. Both methods performed very similar 389 producing complementary mRNA detection patterns (Suppl. Fig. S7). Hand-cutting of clubroot 390 tissue was the best and most reliable method to prepare samples for microscopy (Suppl. Note 391 S1), which resulted in slightly uneven samples. Because of this, CLSM was the method of optical slices these dots often accumulated to larger signal hotspots (Fig 2c, Suppl. Fig. S9 e, 400 f). In maximum projections dots were sometimes well separated (Fig 2c), while in other cells 401 the signal was less defined appearing "blurred" and filling most of the cell without clear 402 structural pattern (Fig 2e), indicating areas with multiple mRNA copies. The controls did not 403 show any comparable signals, only a very weak background of autofluorescence was detected 404 (Suppl. Fig. S5 c-e, Fig. S9 a-b).

407
To evaluate whether it is possible to detect host transcripts within the same samples that host 408 the phytomyxids, two mRNAs were chosen and analysed. which were infected with P. brassicae (Fig. 2 i-j). The signal for MEX1 was not present in cells 417 filled with spores and cells that did not contain amyloplasts (Suppl. Fig. S10). Similar to the phytomyxid Maullinia ectocarpii (Fig. 2 f-h, Suppl. Fig. S11 g-i). Likewise, signals were 434 detected in M. ectocarpii infected M. pyrifera cells (Suppl. Fig. S11 a-f). In both cases, signals 435 were dot-like, yet much more confined -than in the plant and phytomyxids tested. bp to fit a relevant number of probes, while this size restriction does not apply to RCA-FISH. 480 We proved here that the two contrastingly complex FISH methods can be used to generate 481 meaningful results (Tab. 2).

482
The mRNA signal pattern differed between plants, brown algae and P. brassicae (Fig. 2). Plant in cells and were similar in size and shape ( Fig. 2i-j, Suppl. Fig. S10). The mRNA pattern in 486 brown algae are comparable to the ones in plants, although signals were somewhat more 487 restricted ( Fig. 2f-h, Suppl. Fig. S11). The mRNA pattern observed in phytomyxids are only 488 comparable when few mRNAs are detected in the plasmodia (e.g. Fig 2b). However, when 489 many mRNAs are present close to each other the individual mRNAs do not resolve and are 490 displayed as larger areas, especially in the maximum projections of the z-stacks (Fig. 2, Fig 3,   491 Video S1). Although phytomyxid plasmodia are similar in size and shape to their host cell ( Fig.   492 3), each plasmodium has to be interpreted as an aggregation of hundreds to thousands small 493 cells with an individual diameter of 3-5 µm and each with its own nucleus (compare e.g. Fig 2b   494 insert, or Fig 2 f-h). So the space which can be populated by mRNAs is much more confined 495 than in the comparably gargantuan host cells. This is visually amplified in maximum projections of z-stacks, because one plasmodium contains more than one layer of cells (one layer per 3-497 5µm).. It is, however, very important to note that these technical and biological limitations are 498 clearly counterbalanced by the amount of biological information that can be gathered. 499 500 501

RCA FISH smFISH
Pros  Signal amplification in theory allows to monitor single copy mRNAs also in samples with a high fluorescent background.  Specificity: Highly specific probes can be designed to distinguish between paralogues or SNPs .  No gene length limitation: short (<500 bp) genes can be analysed.  Modular: Possibility of combination with in situ proximity ligation assays to collect information on mRNA-protein interactions or post-translational modification (not tested in the scope of this paper).
 Signal amplification in theory allows to monitor single copy mRNAs in samples with a high fluorescent background. give good results after more than one year. Another reason for decreasing signals can be 517 formaldehyde mediated covalent interlinking of mRNAs and the RNA-binding proteins which 518 are responsible for the transport of the RNA to the site of translation (Foley et al., 2017; probes to the sites of interest (Ding et al., 2013). Because of this, we recommend not to store 521 the samples in the formaldehyde containing fixative, but to move them to pure ethanol for long 522 term storage. RNA-binding proteins and mRNA secondary structure can impact on the success 523 of the detection method without the fixation bias mentioned above, because both can limit the 524 accessibility of the target site for probes (Foley et al., 2017). 525 Also, the permeability of cell walls and plant tissue as a whole is a constraint for FISH-based 526 methods. In this study, RCA-FISH was not adaptable to study interactions in filamentous algae, is linked to the expression of a putative effector that alters the host defence response.

555
Using FISH we clearly demonstrate, that PbBSMT mRNAs start to increase when P. brassicae 556 transitions from plasmodial growth to resting spore formation. The detected mRNAs peaked 557 when young, immature resting spores became recognisable (Fig. 3). These results again confirm