Homologues of key circadian clock genes present in Verticillium dahliae do not direct circadian programs of development or mRNA abundance

Many organisms harbour circadian clocks that promote their adaptation to the rhythmic environment. While a broad knowledge of the molecular mechanism of circadian clocks has been gained through the fungal model Neurospora crassa, little is known about circadian clocks in other fungi. N. crassa belongs to the same class as many important plant pathogens including the vascular wilt fungus Verticillium dahliae. We identified homologues of N. crassa clock proteins in V. dahliae, which showed high conservation in key protein domains. However, no evidence for an endogenous, free-running and entrainable rhythm was observed in the daily formation of conidia and microsclerotia. In N. crassa the frequency (frq) gene encodes a central clock protein expressed rhythmically and in response to light. In contrast, expression of Vdfrq is not light-regulated. Temporal gene expression profiling over 48 hours in constant darkness and temperature revealed no circadian expression of key clock genes. Furthermore, RNA-seq over a 24 h time-course revealed no robust oscillations of RNA in constant darkness. Comparison of gene expression between wild-type V. dahliae and a ΔVdfrq mutant showed that genes involved in metabolism, transport and redox processes are mis-regulated in the absence of Vdfrq. In addition, VdΔfrq mutants display growth defects and reduced pathogenicity in a strain dependent manner. Our data indicate that if a circadian clock exists in Verticillium, it is based on alternative mechanisms such as post-transcriptional interactions of VdFRQ and the WC proteins or the components of a FRQ-less oscillator. Alternatively, it could be that whilst the original functions of the clock proteins have been maintained, in this species the interactions that generate rhythmicity have been lost or are only triggered when specific environmental conditions are met. The presence of conserved clock genes in genomes should not be taken as definitive evidence of circadian function. Author summary Circadian clocks are used by organisms to orchestrate the activity of cellular processes such that they occur at an optimal time of day. Research carried out in the filamentous fungus Neurospora crassa has revealed a huge amount of information about the components its circadian clock, its interactions with the environment and how it drives cellular biochemistry and physiology. Although homologues of the Neurospora clock genes are present in a number of fungi, functional clocks have been demonstrated in a just a handful. Importantly, a link between the circadian clock of the plant pathogen Botrytis cinerea and virulence has recently been reported. We report that another significant plant pathogen, Verticillium dahliae, contains well-conserved homologues of all key clock genes. We find that diurnal development of conidia and microsclerotia is not influenced by a circadian clock. Furthermore, in a constant environment we find no evidence of rhythmic transcript accumulation. However, deletion of the central clock component results in altered growth and reduced virulence. This led us to question the role of clock genes in Verticillium. We are forced to consider that in this species the interactions that generate rhythmicity have been lost, are generated purely via post-transcriptional modification of clock proteins, are only triggered when specific environmental conditions are met or never evolved.

The aim of this work was to determine whether a circadian clock is present in V. dahliae. As 128 previously described [36], under 24 hour light/dark cycles V. dahliae displays concentric rings 129 of conidia and microsclerotia. We report that this phenotype is directly driven by external cues, motif discovery, DREME [44] and GLAM2 were used for ungapped and gapped motifs, 178 respectively. Motif enrichment analysis was carried out with the AME program. A manual 179 screen of consensus LRE motifs described in [15,45]  Strains of V. dahliae (S1 Table)  DESeq2 package in R. A grouping command was used to assess differentially expressed genes 280 (DEG) between conditions (L or D) and strains (Vd12253, Vd12008 and ΔVdfrq_12253). The 281 false discovery rate (FDR) cut-off was set to 0.05. Genes were considered to be significantly 282 differentially expressed when p-value < 0.05 and presented more than 1-log2 fold change 283 For rhythmic expression analysis, the raw counts of the time-course samples were normalized 285 to fragments per kilobase of transcript per million mapped fragments (FPKM). The time-course 286 of each individual strain was analyzed using JTK-CYCLE, looking for rhythms of 24 h [52]. 287 Genes with a p-value < 0.05 and a false discovery rate of q < 0.1 were considered rhythmic. 288 Principal component analysis (PCA) plot and samples distances plot used rlog-transformed 289 read counts, and were carried out using R. Gene ontology (GO) terms were retrieved from 290 GO.db in R. Gene enrichment analysis for GO terms was performed using topGO in R with 291 Fisher's exact test to retrieve significantly enriched processes of the DEG. The analysis of 292 secondary metabolite biosynthetic gene clusters in V. dahliae was performed using 293 antiSMASH [55]. 294 295

Construction of V. dahliae frq replacement cassette 296
The 5' and 3' flanking regions of the Vdfrq locus were amplified from genomic DNA using 297 primer pairs HRFrq1-F/R and HRFrq2-F/R respectively (S2 Table). Core  spore colonies were obtained by plating 100 μL of the culture on PDA with 50 μg/ml 321 hygromycin B. PCR was carried out on selected single-spore mutants to confirm insertion of 322 hygromicin using the primers TestHygr-F/R and FrqUS_Hygr-F/R, and lack of product from 323 deleted gene using the primers TestFrq-F/R (S1 Fig.)

Arabidopsis thaliana and strawberry in vitro propagation 328
For pathogenicity tests, seeds of Arabidopsis thaliana ecotype Columbia (Col), were surface-329 sterilized by sequential immersion in 70% ethanol for 1 min and 10% (v/v) bleach containing 330 0.1% (v/v) Tween-20 for 5 minutes with gentle shaking. Seeds were then washed with 331 autoclaved distilled water for 5 minutes five times, re-suspended in sterile 0.1% agarose, and 332 subsequently stratified at 4 o C over a period of 2-3 days in darkness [59]. 5 to 6 seeds were 333 pipetted into 120 x 120 mm square petri dishes (Thermo Fisher Scientific) containing half-  The genomes of 25 Sordariomycete, Dothideomycete and Leotiomycete species were searched 378 for homologues of frq, wc-1, wc-2, frh, fwd-1 and vvd. The predicted proteins from these 379 genomes were clustered and proteins in the same orthogroup cluster as the N. crassa genes 380 were considered orthologous genes. Orthologues of all six clock oscillator genes are present in 381 most species of Sordariomycetes tested in this study, including important plant-pathogenic fungi such as Glomerella graminicola and Neonectria ditissima (Fig 1A) Venturia inaequalis, but present in the wheat pathogen Zymoseptoria tritici that contains 391 homologues of all the clock genes (Fig 1A). The genome of the Ustilaginomycetes Ustilago 392 maydis, on the other hand, does not harbour an orthologue of frq, in agreement with the findings 393 in [28]. 394 Alignments of core clock proteins between N. crassa and V. dahliae revealed sequence 395 identities greater than 43% and query coverages greater than 39%. Consistent with previous 396 reports [29], V. dahliae FRQ is the least conserved (43.82% identity), and FRH is the best 397 conserved (68.69% identity) of the core clock proteins. Clock proteins are strongly conserved 398 at the domain level (Fig 1B). Although the FRQ protein alignment created between N. crassa 399 and V. dahliae sequences shows moderate conservation overall, a number of important regions hour LD cycles at constant temperature, hyphal growth occurs under both light and darkness, 453 however conidiation from hyphae is induced after a period of light (Fig 2A and B). conditions V. dahliae cultures contain conidia and microsclerotia but they are not produced in 457 a ring fashion (Fig 2B). To test if development is controlled by a circadian clock we 458 investigated whether the morphological rhythms found in this pathogen are endogenous and 459 can be entrained. We found that V. dahliae cultures transferred to DD after an entrainment 460 period of 12:12 LD for 14 days at 22 °C cease to produce rings of conidia and microsclerotia. 461 In 12:12 temperature cycles of 20 °C and 28 °C in DD or LL the banding pattern is synchronized 462 to the temperature oscillations, and ceases when cultures are transferred to constant 463 temperature (Fig 2C). Thus, the developmental rhythms are induced by light and temperature 464 cycles, rather than being self-sustainable. 465 Circadian clocks are able to entrain to cyclical cues in the environment and in all organisms 466 studied to-date, light experienced late in the subjective day causes phase delays of clock time 467 whilst light experienced late in the subjective night causes phase advances. This affects 468 rhythms that are clock-controlled such that in changing day-lengths clock-controlled outputs 469 occur at the correct time of day [3]. This ability to reset also allows circadian clocks to entrain 470 to environmental cycles shorter and longer than 24 hours (T cycles). A phenomenon known as 471 frequency demultiplication emerges when organisms are exposed to very short or long periods 472 (e.g. 6:6 or 24:24 LD or temperature cycles). That is, they display a 24-hour rhythm as if the 473 oscillation is entrained to a 12:12 cycle [3]. In contrast, if a rhythm is simply a direct response 474 to external cues, it assumes the periodicity of the driving light or temperature cycle [3]. In V. 475 dahliae, cultures grown under 6:6 LD and 6:6 temperature cycles show narrow conidial and 476 microsclerotia bands, whereas cultures under 28:28 LD and 28:28 temperature cycles form 477 widely spaced bands (Fig 2D). Thus, the developmental rhythm shows no evidence of 478 frequency demultiplication. Instead, conidial and microsclerotia production follows each 479 light/dark and temperature transition. In contrast, exposed to the same conditions, robust oscillation of N. crassa frq is seen (Fig 3A) 499 and, in agreement with previous studies [79], frq expression peaked at 12 h and 36 h after 500 transfer to DD (subjective morning). Another feature of circadian rhythms is that they 501 anticipate cyclic changes in the environment. Therefore, we assessed the expression of several 502 clock genes (frq, wc-1, wc-2 and vvd) in a 12:12 LD cycle, with high time-point resolution 503 before and after "lights on" (Fig 3B). Vdfrq transcript levels under DD seemed to anticipate 504 the transition to light as the expression slightly increased from D10 to D11.5, albeit with less than a 1 log2 fold change. However, Vdfrq transcript levels do not exhibit significant rhythmic 506 oscillation (p-value=1) and drop after the first time-point in the light. In contrast, the expression 507 of Vdwc-1, Vdwc-2 and Vdvvd do not anticipate the dark/light transition. There is a slight 508 increase in the expression of Vdwc-1 and Vdwc-2 after 2 hours in the light, whereas Vdvvd is 509 highly expressed after the dark to light transition, and drops to basal levels at a later time-points 510 possibly due to photoadaptation [80] (Fig 3B). In conclusion, there is no strong evidence for 511 anticipatory behaviour that could indicate light entrainment of a V. dahliae circadian clock, nor 512 is there significant rhythmic gene expression in LD. Similarly, there is no significant difference 513 in the expression of Vdfrq, Vdwc-1, Vdwc-2 or Vdccg-16 before or after the transition between 514 a high temperature (HT) period of 28 °C or a low temperature (LT) period of 20 °C (Fig 3C). that V. dahliae frq expression is not induced by pulses of light (Fig 3D), whereas, as expected, 556 N. crassa frq transcript increases 10-fold. Furthermore, in contrast to N. crassa [74], the 557 expression of Vdwc-1 also lacks induction by light (Fig 3D). In contrast, orthologues of the 558 light-inducible N. crassa genes vvd and cry are both up-regulated. Vdcry mRNA increases 14-559 fold and Vdvvd mRNA 5-fold in response to light (Fig 3D). [52]. The parameters were set to look for rhythms with a period of 24 hours. An important 568 point to bear in mind is that JTK-cycle duplicates the 24-hour time series and then analyses the 569 data for rhythmicity. Thus, as well as transcripts that reach their peak or nadir at DD12 and 570 DD18, transcripts that decrease or increase with time after the light to dark transfer will be 571 identified as rhythmic. 572 In V. dahliae 12253, significant oscillation in expression is observed for 568 genes 573 (approximately 4.9 % of the genome; p < 0.05) although the q value (the minimum false 574 discovery rate at which the test is considered significant) ranges from 0.2 to 0.8. The amplitude of the cycle is lower than 10 in 432 genes, which suggests a very weak oscillation. In V. dahliae 576 12008, 884 genes are identified with rhythmic expression, with a p-value < 0.05 and a q value 577 ranging from 0.1 to 0.5. In this isolate vvd, cry-dash and putative clock-controlled genes ccg-578 9 and ccg-16 are significantly rhythmic, with amplitudes ranging from 1.4 to 21.5. However, 579 most oscillations are weak with amplitudes lower than 10 in 83% of the rhythmically expressed 580 genes. Of the 568 and 884 rhythmic genes identified from V. dahliae 12253 and 12008, only 581 34 genes are rhythmic in both strains (S8 Fig). For those genes, the cycle amplitude is low with 582 q values higher than 0.1 (S3 Table) and several (Chr1g02750, Chr1g18780, Chr2g02830, 583  (Fig 4). In order to validate the RNA-seq results, 589 V. dahliae 12253 and 12008 samples harvested every 4 hours over a 48 hour time-course were 590 analysed by qRT-PCR (Fig 3A). In both strains, cry-dash and vvd were arrhythmic. This result 591 highlights the possible misinterpretation of transcripts as rhythmic by JTK that are simply 592 falling or rising after the light to dark transfer.  (Fig 5A). Despite the variability 601 among replicates we observed a reduction in total colony growth for both Δfrq mutants 602 (Δfrq_12253 and Δfrq_12008) with respect to the WT (12253 and 12008) strains (p-value < 603 0.01, p-value < 0.01, respectively). The reduction in daily growth was independent of lighting 604 conditions (Fig 5A) but dependent on the nutritional composition of the culture medium (S9 605 Δfrq_12253 grown in either light or dark were involved in redox processes, heme oxidation, 632 circadian rhythms and glutamate biosynthetic processes. Additionally, genes that are down-633 regulated only in the light are functionally enriched for phosphate ion transport, nitrate 634 assimilation transport, superoxide anion generation, one-carbon metabolic processes as well as 635 pathogenesis processes (S4 Table). 636 637

V. dahliae frq regulates the expression of photoreceptor, TF-and SM-encoding genes. 638
In Neurospora, the transcription factor complex of the clock oscillator machinery activates the 639 transcription of many other TFs which results in a gene regulatory cascade [84]. As a result, the 640 circadian clock modulates the expression of genes involved in many processes, especially those 641 involved in physiology and metabolism [21]. Therefore, we hypothesized that disruption of 642 Vdfrq would affect the expression of photoreceptors, transcription factor-encoding genes and 643 genes involved in secondary metabolism.  Table). Interestingly, when exposed 649 to light vvd is significantly up-regulated in the Δfrq_12253 mutant, indicating that VdFRQ has 650 a negative effect on vvd transcription. In the case of cry-dash gene, its expression is light-651 regulated in both the WT and Δfrq_12253, although the absolute expression level in the Vdfrq 652 mutant is significantly lower than in the WT. The expression profile of the rgs-lov, cry-1, phy, 653 phr and nop-1 genes does not change in the absence of Vdfrq. 654 We found that the absence of Vdfrq also affects the expression of several TF-encoding genes. 655 A total of 50 genes containing functional annotations associated with TF exhibited differences 656 in expression due to the lighting conditions or strain background. Most of the differences in 657 expression were due to the Vdfrq mutation. 26 TF-encoding genes were up-regulated, and 8 658 TF-encoding genes were down-regulated in Δfrq_12253 regardless of the light conditions. 659 However, there were 6 TF-encoding genes that were no longer light-induced or light-repressed 660 in Δfrq_12253 mutant strain (S5 Table). Vdfrq. Interestingly, the light-induced expression of these genes was compromised in 665 Δfrq_12253 (S6 Table) Table). Therefore, Vdfrq is crucial for the regulation of expression of secondary-metabolism-encoding genes in V. isolates were isolated from UK strawberries and fall within the VC group subclade II-2, 12008 683 being a highly virulent isolate and 12253 being a moderately virulent isolate. The infected 684 seedlings were incubated in a 12:12 LD cycle for 28 days. Symptoms were visually rated at 0, 685 7, 14, 21 and 28 dpi on a scale of 1 to 9, in which 1 was equal to no symptoms and 9 equalled 686 a dead plant. In A. thaliana seedlings infected with the wild-type strain WT_12253 present 687 symptoms in up to 75% of the leaves after 21 days of inoculation (Fig 5C). At the same time 688 post inoculation plants infected with Δfrq_12253 show symptoms of wilt on 20% of leaves. 689 The difference is more obvious at 28 dpi, when most plants infected with WT_12253 are dead 690 whilst Δfrq_12253 infected plants display slight chlorotic symptoms in several outer leaves. 691 The AUDPC confirmed a significant difference in pathogenicity between Δfrq_12253 and 692 WT_12253 strains (p-value < 0.01). Contrary to this observation, WT_12008 and Δfrq_12008 693 do not present differences in virulence (Fig 5D). Similar results are obtained from 694 pathogenicity tests on a susceptible strawberry cultivar (Hapil). After 6 weeks of inoculation, 695 plantlets infected with the Δfrq_12253 strain show fewer disease symptoms than the 696 WT_12253 strain (Fig 5E), whereas the WT_12008 and Δfrq_12008 strains do not show 697 differences in the ability to cause disease (Fig 5F). These results indicate that pathogenicity of the Vdfrq mutant is impaired in an isolate-dependent manner, with isolates with greater degrees 699 of pathogenicity showing non-significant differences. has an effect on the light response. One reason for this could be that in ΔVdfrq expression of 808 Vdwc-1 is down-regulated. This prompts speculation that a Vdwc-1 deletion would also affect 809 pathogenicity. To summarise, our data reveal large changes in gene expression, altered growth 810 and pathogenicity in the Vdfrq deletion mutant. Whether or not these phenotypes result from 811 VdFRQ functioning outwith a circadian clock cannot at present be ascertained. 812 With regard to the existence of a circadian clock in V. dahliae our results suggest three 813 possibilities; i) the clock is absent, ii) the clock is post-transcriptional and constitutive gene 814 expression leads to oscillation at the protein level, iii) the clock is only active during specific 815 developmental stages and /or specific conditions e.g. the clock is activated when a host is 816 detected and is only functional in planta. As V. dahliae infects and moves through host tissue 817 it is likely that an ability to anticipate time-of-day changes in host immunity would be 818

beneficial. 819
At least three circumstances can be envisaged where circadian rhythmicity might be absent. 820 The first is when an organism is always ready to respond to the rhythmic environment. It has 821 been reported that despite the presence of homologues of most clock genes in Picea abies