A stress-responsive p38 signaling axis in choanoflagellates

Animal kinases regulate cellular responses to environmental stimuli, including cell differentiation, migration, survival, and response to stress, but the ancestry of these functions is poorly understood. Choanoflagellates, the closest living relatives of animals, encode homologs of diverse animal kinases and have emerged as model organisms for reconstructing animal origins. However, efforts to study kinase signaling in choanoflagellates have been constrained by the limitations of currently available genetic tools. Here we demonstrate that small molecule approaches provide a complementary and scalable approach for studying kinase function in choanoflagellates. To study the physiological roles of choanoflagellate kinases, we established two high-throughput platforms to screen the model choanoflagellate Salpingoeca rosetta with a curated library of human kinase inhibitors. We identified 95 diverse kinase inhibitors that disrupt S. rosetta cell proliferation. By exploring structure-activity relationships of one inhibitor, sorafenib, we identified a p38 kinase as a regulator of heat and oxidative stress in S. rosetta. This finding indicates a conserved p38 function between choanoflagellates, animals, and fungi. Moreover, this study demonstrates that existing kinase inhibitors can serve as powerful tools to examine the ancestral roles of kinases that regulate modern animal development.


Introduction
Phenotypic screens with libraries of small molecules have revolutionized cell biology by providing chemical tools to study protein function [1][2][3][4] .Because aberrant kinase activity leads to human disease 5,6 , many tools have been developed to inhibit kinase activity and detect protein phosphorylation.Small molecules that target the kinase active site coupled with assays of kinase inhibition have uncovered effective therapeutic strategies to counter the functions of misregulated kinases, including inhibition of aberrant cell growth and proliferation caused by oncogenic kinases 5,7 .
We sought to test whether kinase-regulated physiology in choanoflagellates, the closest living relatives of animals, could be revealed by kinase inhibitors.Choanoflagellates possess homologs of diverse animal kinases (Figure S1) [8][9][10] and, due to their phylogenetic placement, are well-suited for studies of the ancestral functions of animal cell signaling proteins 11,12 .
Although targeted genome editing has been established in choanoflagellates 13 , we inferred that the unique strengths of small molecules, including precise temporal control and discrete inhibition of enzymatic activity 14 , might more readily reveal the physiological relevance of kinases, including those with essential functions, in choanoflagellates.Indeed, a previous study showed that two broad-spectrum kinase inhibitors disrupt cell proliferation in the choanoflagellate Monosiga brevicollis 15 , but did not demonstrate whether kinases were directly targeted or identify specific pathways regulated by kinase signaling.
Using a library of well-characterized kinase inhibitors that vary in their human kinase inhibition profile, we treated cultures of Salpingoeca rosetta, a model choanoflagellate, in a multiwell format.We found that treatment of S. rosetta cultures with a set of kinase inhibitors disrupted cell proliferation and led to global inhibition of S. rosetta phosphotyrosine signaling.Using one of these inhibitors, sorafenib, as a probe and inactive analogs as negative controls, we found that an S. rosetta p38 kinase homolog is activated by environmental stressors and signals downstream of sorafenib-inhibited kinases.

Screening of a human kinase inhibitor library reveals small molecules that inhibit S. rosetta kinase signaling and cell proliferation
To investigate whether kinase activity regulates S. rosetta cell proliferation, we treated S. rosetta cultures with characterized human kinase inhibitors.Because the highest conservation between choanoflagellate and human kinases occurs in the kinase domain [8][9][10] , we focused on kinase inhibitors that bind in the kinase active site.As a proof-of-concept, we first assayed staurosporine, a well-characterized broad-spectrum kinase inhibitor and inducer of cell death in diverse organisms 16,17 .Our initial screen used flow cytometry to measure the density of S. rosetta cells individually treated with staurosporine in multiwell plates.Staurosporine significantly and reproducibly reduced S. rosetta cell density (Figure S2A-C) and tyrosine phosphorylation (Figure S2D) in a dose-dependent manner.
After validating our flow cytometry pipeline with staurosporine, we expanded our study to screen 1255 inhibitors of diverse human kinases (Figure S1, Table S1).We screened S. rosetta cultures with each of the molecules in the library at 10 µM and measured cell density at a 24hour endpoint (Figure 1A, Figure 1B, Table S1).This screen revealed 44 compounds (3.5% of the library) (Figure 1A and Figure S3) that significantly decreased S. rosetta cell counts compared to DMSO controls.
As a secondary assay, we pursued an imaging-based workflow to measure cell number.
Although S. rosetta cell counts could be measured by flow cytometry without staining reagents, our co-culture system and treatment paradigm presented two potential sources of inaccuracy: compound aggregation due to low solubility in choanoflagellate media and choanoflagellatesized clumps of bacterial biofilm.Therefore, we developed an imaging pipeline to enumerate S. rosetta cells by segmenting fixed-cell immunofluorescence micrographs at a 48-hour endpoint.
Our imaging pipeline was able to distinguish wells with staurosporine-treated cells from DMSO controls (Figure S2E, F) with comparable z' standard statistics to flow cytometry (Figure S2B, S2E) 18 .This orthogonal approach identified 22 compounds (1.8% of the library) that overlapped with the flow cytometry screen and 51 additional molecules (4.4% of the library) that inhibited S. rosetta cell proliferation but were not identified in our flow cytometry screen (Figure 1A, Figure 1C, Figure S2G, Figure S3).In total, both screens identified 95 ATP-competitive inhibitors of human protein and lipid kinases (Figure S4A-C) that ranged in selectivity (Figure S4D).Our strategy of using two complementary screening approaches uncovered human kinase inhibitors that interfere with S. rosetta cell proliferation.Choanoflagellates are predicted to express kinases that regulate animal cell growth, including mitotic kinases 19 , the serinethreonine kinase Akt 20 , and a diverse set of tyrosine kinases 8 .We identified GSK461364 and Volasertib, inhibitors of polo-like kinase 1 (PLK1), a mitotic kinase, in both screens (Figure S3).
In addition, both screens identified diverse inhibitors of human Akt and tyrosine kinases (Figure S3).We also identified inhibitors of S. rosetta cell proliferation that disrupt human kinase signaling indirectly, i.e. without binding to a kinase (Figure S3).For example, genistein, a natural product that indirectly alters kinase activity 21,22 , inhibited S. rosetta cell proliferation and was previously shown to inhibit the growth of the choanoflagellate M. brevicollis 15 .

Some inhibitors of S. rosetta cell proliferation also disrupt S. rosetta phosphotyrosine signaling
After identifying human tyrosine kinase inhibitors that inhibited S. rosetta cell proliferation in our screens, we used biochemical methods to determine if the phenotype observed was correlated with inhibition of S. rosetta kinase activity.Widely available commercial phosphospecific antibodies that distinguish phosphorylated amino acids can reveal kinase activity in diverse organisms 15,[23][24][25][26] and have been used to detect tyrosine phosphorylation by heterologously expressed choanoflagellate kinases [27][28][29][30] .We triaged our set of 95 identified proliferation inhibitors to focus on those compounds that inhibit human tyrosine kinases (e.g.tyrosine kinase inhibitors, multi-class targeting kinase inhibitors) (Figure S3), as opposed to those that inhibit serine or threonine phosphorylation, due to the relative lack of specificity of commercially-available phosphoserine and phosphothreonine antibodies [24][25][26] .
Of the four inhibitors tested, S. rosetta cultures were most sensitive to sorafenib and glesatinib.Although 1 µM masitinib and PP121 were sufficient to reduce S. rosetta cell proliferation over the first 40 hours of treatment (Figure 2A), masitinib-and PP121-treated cultures recovered within 85 hours.In contrast, sorafenib and glesatinib inhibited cell proliferation throughout the 85-hour growth experiment at 1 µM (Figure 2A) and decreased cell density at 1 µM and 10 µM (Figure 2B).Glesatinib treatment induced cell lysis (Movie S1), whereas sorafenib induced cell body elongation (Movies S2-S3), in comparison to DMSO control (Movie S4).Importantly, treatment of S. rosetta cultures with 1 µM sorafenib or glesatinib led to a global decrease in phosphotyrosine signal, while treatment with 1 µM masitinib or PP121 did not decrease phosphotyrosine levels as detected by western blot (Figure 2C).These findings showed that some kinase inhibitors, including sorafenib and glesatinib, could disrupt both S. rosetta cell proliferation and tyrosine kinase signaling.To further investigate patterns among human kinase inhibitors that showed this effect and identify kinases that may be relevant for the observed inhibition, we tested a panel of 17 human TK inhibitors (Figure S6) that share overlapping kinase targets with sorafenib and glesatinib.Of the 17 compounds tested, treatment with four additional small molecules (regorafenib, AD80, milciclib, and vemurafenib) led to a global decrease in phosphotyrosine staining (Figure S6A, "*") but not phosphoserine and phosphothreonine staining (Figure S7) as detected by western blot.These four inhibitors mildly reduced the rate of cell proliferation (Figure S6B, Figure S8).Other tyrosine kinase inhibitors inhibited S. rosetta cell proliferation but did not inhibit phosphotyrosine signaling, including PP2 (Figure S6C, Figure S8), consistent with prior findings in M. brevicollis 15 .Together, these observations provide independent support for the hypothesis that select kinases regulate cell proliferation in choanoflagellates.

Sorafenib binds to S. rosetta p38 kinase
To identify specific S. rosetta kinases whose activity might regulate S. rosetta cell proliferation, we focused on sorafenib, an inhibitor with well-characterized structure-activity relationships in animals.We started with a kinase-enrichment strategy previously used in animals and plants to enrich kinases out of prepared lysates 40,41 .We used ActivX ATP and ADP probes 40,41 to covalently enrich for kinases and high-affinity kinase-binders present within S. rosetta lysates after pretreatment with vehicle (DMSO) or sorafenib.
Because sorafenib and ActivX probes competitively bind to kinase active sites, kinases that were more often enriched with DMSO pretreatment and less enriched with sorafenib pretreatment were likely sorafenib binders.We quantified absolute protein abundance 42 and focused our attention on kinases that were present in both DMSO and sorafenib pretreated samples but were enriched in multiple DMSO replicates, represented by higher ProteinLynx Global Server (PLGS) protein scores 42 .Through this approach, we identified a predicted S. rosetta p38 kinase (Figure S9A) that bound 10-fold less well to ActivX probe in the presence of sorafenib (Figure 3, Table S2).Hereafter, we refer to this protein as Sr-p38.
ActivX probes and other unbiased approaches have previously identified human p38 kinases as primary targets of sorafenib 43,44 .In animals, sorafenib preferentially binds kinases with threonine gatekeeper residues 44 , including tyrosine kinases and two of four vertebrate p38 kinase paralogs with threonine gatekeepers 44 (Figure S9A).Sr-p38, contains a threonine gatekeeper (Figure S9A) and lysine residues necessary for ActivX probe binding.Because sorafenib displaced ActivX probe binding to Sr-p38, we infer that sorafenib binds to Sr-p38.

Kinases upstream of p38 kinase regulate cell proliferation in S. rosetta
To uncover the in vivo mechanism by which sorafenib inhibited S. rosetta cell proliferation, we compared the response of S. rosetta wild-type and p38-mutant cell lines to treatment with a panel of p38-selective human kinase inhibitors.In a previous study, two of four human p38 kinase paralogs with threonine gatekeepers, p38a and p38b bound less to skepinone-L, a p38-selective human kinase inhibitor, and sorafenib when the p38 threonine gatekeeper was mutated to methionine.We reasoned that if the sensitivity to p38 inhibitor binding was conserved in Sr-p38, and if Sr-p38 directly regulated cell proliferation in S. rosetta, Sr-p38 T110M mutant cell lines would lose sensitivity to sorafenib, skepinone-L and BIRB 796, another p38-selective human kinase inhibitor that more potently inhibits human p38a and p38b.
Sorafenib, skepinone-L and BIRB 796 reduced wild-type S. rosetta cell proliferation at 1 µM (Figure 4A) and almost entirely abrogated cell proliferation at 10 µM during an 80-hour growth course (Figure 4A).In addition, S. rosetta cells treated with skepinone-L showed a similar cell elongation response to sorafenib treatment (Movies S2, S3, S5).We generated two independent Sr-p38 T110M cell lines by isolating two clones from a population of cells that underwent CRISPR-mediated gene editing 13 to mutate ACA to ATG at codon 110 of Sr-p38 (Figure 4B).Co-editing for cycloheximide (CHX) resistance was used to enrich for clones that were nucleofected with the Cas9 ribonucleoprotein and grew in the presence of cycloheximide, which is toxic to wild-type S. rosetta cells 13 .Both Sr-p38 T110M cell lines, CHX Sr-p38 T110M-1 and CHX Sr-p38 T110M-2 , grew similarly to cycloheximide-resistant controls, CHX 1 and CHX 2 , (Figure 4C) and were not desensitized to sorafenib, skepinone-L or BIRB 796 treatment (Figure 4D).Thus, we concluded that sorafenib, in addition to binding directly to Sr-p38, independently inhibits kinases upstream of Sr-p38 and thereby decreases S. rosetta cell proliferation.Future studies will be required to identify additional binding targets of sorafenib.

Sorafenib inhibits the stress-responsive p38 kinase signaling axis conserved in S. rosetta
Because we did not find a direct connection between inhibition of Sr-p38 activity and inhibition of S. rosetta cell proliferation, we sought to identify an acute physiological role for Sr-p38 that could be altered by sorafenib treatment.Phylogenetic analyses place the origin of p38 in the last common ancestor of animals and fungi 45 .Environmental stressors activate p38 kinases in animals and fungi [45][46][47][48][49][50][51][52] and p38 kinase is present in diverse choanoflagellates (Figure S9B).However, the roles of p38 kinase in choanoflagellate biology are unknown.Although a previous study identified a nutrient-sensitive protein in M. brevicollis with a molecular weight similar to p38 kinases 15 , the identity and function of this protein were not directly studied.
We wondered if stressors relevant to choanoflagellates would activate Sr-p38 signaling.Some choanoflagellates, including S. rosetta, live in sun-lit water zones that undergo daily and yearly fluctuations in temperature and nutrients [53][54][55] .To investigate if Sr-p38 is activated in response to environmental stressors, we exposed S. rosetta cultures to heat shock and oxidative stress.When S. rosetta cells cultured at ambient temperature were subjected to heat shock and oxidative stress, we observed an increase in Sr-p38 phosphorylation (Figure 5A, B).
In animals, upstream dual-specificity kinases (MAP2Ks) activate p38 by phosphorylating the threonine and tyrosine of the "TGY" motif 50,56 , and this phosphorylation is recognized by several phospho-specific antibodies (Figure S9C).Within 30 minutes of heat shock at 37°C, the phospho-p38 signal from at least two protein species increased relative to pre-treatment (Figure 5A).We also observed an increase in signal for at least one protein using the phospho-specific p38 antibody on cell lysates that were treated with 0.5M hydrogen peroxide (Figure 5B).
Because Sr-p38 contains the diagnostic "TGY" motif, we infer that this antibody recognizes Sr-p38.Thus, the connection between environmental stress and the activation of p38 kinase appears to be conserved between yeast, animals, and choanoflagellates 45 .
Previous studies have connected phosphotyrosine signaling and p38 kinase activation in animals in response to multiple stimuli (e.g.growth factors, ultraviolet light) 50,57 .Sorafenib blocks this signaling axis by inhibiting p38 and upstream kinases (e.g.tyrosine kinases, dualspecificity kinases) 44,58 .To examine if the signaling axis between upstream kinases and p38 is conserved in S. rosetta, we tested whether sorafenib could reduce the observed phosphorylation of Sr-p38 in S. rosetta cultures subjected to heat shock.As a control, we used APS6-46, a sorafenib analog that shares sorafenib's core structure but has modifications that make it too large to bind to most sorafenib kinase targets 59 .APS6-46 does not inhibit S. rosetta tyrosine phosphorylation or cell division (Figure S10), and a previous study found that similar sorafenib analogs could control for off-target inhibition of non-kinase proteins by sorafenib 59 .Sr-p38 signaling was not activated in S. rosetta cultures pretreated with sorafenib, whereas cultures pretreated with APS6-46 (Figure 5C) or human p38 kinase inhibitors (Figure S11) retained Sr-p38 phosphorylation comparable to the DMSO control.Because sorafenib treatment blocked Sr-p38 activation and inhibited phosphotyrosine signaling, we infer that sorafenib blocks Sr-p38 signal transduction through inhibition of upstream kinases that transduce the heat and oxidative stress response in S. rosetta (Figure 5D).

Discussion
To investigate the relevance of kinase signaling in choanoflagellate physiology, we have established two high-throughput phenotypic screens of cells treated with a small molecule library.By treating S. rosetta cultures with validated human kinase inhibitors, we uncovered molecules that revealed the physiological relevance of kinases as regulators of S. rosetta cell proliferation.Moreover, we identified biologically relevant environmental stressors that activate p38 kinase signaling in S. rosetta.
Until recently, the functions of stress-responsive kinases, including p38, JNK, and Hog1, were only characterized in animals and fungi 45 .However, this family of kinases evolved before the divergence of animals and fungi and expanded in choanoflagellates, which encode at least two paralogs 45 .Our discovery that S. rosetta phosphorylates proteins in response to heat and oxidative shock (Figure 5A, B) demonstrates that choanoflagellates undergo stress-responsive signaling.Because the phosphorylation of these stress-activated proteins can be inhibited by sorafenib (Figure 5C), a known binder of animal p38 44 and Sr-p38 (this study), we infer that Sr-p38 is an S. rosetta stress-responsive kinase.
Our approach of using sorafenib to uncover the role of Sr-p38 and sorafenib-targeted kinases in S. rosetta allowed us to bypass potential limitations with genetic approaches.
Vertebrates express four p38 paralogs; the p38ɑ knockout is embryonic lethal, whereas knockouts for the other p38 paralogs are viable 50 .S. rosetta is predicted to encode at least two stress-responsive kinases, including Sr-p38 and EGD82769 (Figure S9), but it is unknown if they have independent functions.We observed multiple bands when subjecting S. rosetta cultures to heat shock (Figure 5A), suggesting either that Sr-p38 exists in multiple isoforms or that the phospho-p38 specific antibody that is selective for the diagnostic p38 "TGY" motif found in Sr-p38 showed cross-reactivity with the EGD82769 "TPY" motif (Figure S9).Sorafenib inhibited phosphorylation of both proteins (either isoforms of Sr-p38 or Sr-p30 and EGD82769) (Figure S9).The insensitivity of Sr-p38 T110M mutant cell lines (Figure 4D) to sorafenib suggests that the small molecule targets additional stress-responsive kinases.
Choanoflagellates have dynamic life histories and diverse kinase families, including p38 kinases, that are found in animals 8,10,19,45,60 .How kinases regulate additional aspects of choanoflagellate physiology, including life history transitions, remains to be investigated.We infer that fast-acting inhibition of kinase activity will be a powerful approach to study the roles of kinases that regulate cell state transitions in choanoflagellates.Because kinase inhibitors allow the enzymatic activity of kinases to be disrupted while preserving kinase localization and scaffolding functions, using small molecules as tools can distinguish whether kinase catalytic activity or other kinase functions are necessary for choanoflagellate development 14,61 .Insight into the roles of individual kinases during the emergence of new cell-cell signaling networks (e.g., receptor tyrosine kinase signaling in holozoans) will be fundamental to understanding the contributions of kinase signaling to the origin of animal multicellularity.

Co-culturing of S. rosetta with the prey bacterium Echinicola pacifica
Choanoflagellates are bacterivores and require prey bacteria that are co-cultured in choanoflagellate media 62 .Echinicola pacifica, a Bacteriodetes bacterium, grown in seawaterbased media enriched with glycerol, yeast extract, and peptone sustains S. rosetta growth 63 .This co-culture of S. rosetta and E. pacifica, publicly available from the American Type Culture Collection (ATCC) as ATCC PRA-390 and also known as "SrEpac" 63 , was used in this study.

High-throughput chemical screening in S. rosetta
To quantify changes in S. rosetta cell proliferation after small molecule treatment, we established a high-throughput screening pipeline.
We first assembled a library of 1255 compounds (Table S1) from commercial (Selleckchem Kinase Inhibitor Library Catalog #L1200 Lot: Z1316458) and academic (Kevan M. Shokat, University of California, San Francisco) sources.98% of molecules in the library were characterized as human kinase inhibitors and 2% were compounds that are cytotoxic to other protists.Most of the kinase inhibitors in the library modulate human kinase activity by binding to the kinase active site and are ATP-competitive.Because we did not know if inhibitors designed to bind to human kinases would bind to choanoflagellate kinase homologs with the same potency or selectivity, we chose inhibitors with a range of selectivity: 75% of the human kinome is targeted by at least one inhibitor in the library, and the library includes inhibitors of all classified human kinase groups (Figure S1).Compounds dissolved at 10 mM in dimethyl sulfoxide (DMSO, Sigma #D8418) were placed into individual wells of 384-well deep well master plates (Corning #3342).We generated deep well stock plates with solutions that could be directly transferred to assay plates.Liquid handling (Agilent V11 Bravo) was used to dilute compound master plates (containing 10 mM compound in 100% DMSO) into deep well stock plates (containing 450 µM compound in 4.5% DMSO).
In a primary screen, S. rosetta cell counts were determined by analysis of acquired flow cytometry events after a 24-hour incubation.Assay plates were generated by plating 2 µL of the deep well stock plates into 384-well assay plates (Thermo Scientific #142761 using the Agilent V11 Bravo).88 µL of SrEpac cultured in high-nutrient media (5% Sea Water Complete) 64 at exponential phase (~9x10 5 cells/mL) was diluted to 2x10 4 cells/mL in high-nutrient media and dispensed into the assay plate (ThermoFisher Mulitdrop™ Combi with long standard dispensing tube cassette #24072677) to treat the SrEpac culture at 10 µM compound and 0.1% DMSO.
After a 24-hour incubation, assay plates were individually placed into an autosampler (BD Biosciences High Throughput Sampler, HTS) running in high-throughput mode.40 µL of each well was mixed twice (at 180 µL/second), and 10 µL of cells from each well were loaded onto a flow cytometer (BD Biosciences LSR II) at 1 µL/second.In between each well, the autosampler needle was washed with 400 µL of sheath fluid (1X phosphate-buffered-saline pH 7.4).Loaded cell samples were acquired on the cytometer with forward scatter (FSC) and side scatter (SSC) parameter voltages set to 538 and 308, respectively.A polygon gate from a DMSO well within a plate (D23 for Plate 1 and C23 for Plates 2-4 due to a shift in the distribution of observed events) was used to analyze events and quantify cell counts for all wells in an individual assay plate using FlowJo v10.8™ (BD Biosciences) (Figure S2A).
In a secondary screen, S. rosetta cell counts were determined by enumerating segmented cell objects from immunofluorescence images taken on an Opera Phenix highcontent imager after a 48-hour incubation.Assay plates were generated by adding 4 µL of compound in the deep well stock plates into 96-well assay plates containing 176 µL of SrEpac in low-nutrient medium (1% Cereal Grass Medium and 1% Sea Water Complete 13 ) using an electronic multichannel pipetter (Rainin E4 Multi Pipette E8-20XLS+).Cells were initially expanded in high-nutrient media containing 4% Cereal Grass Medium and 4% Sea Water Complete 13 and at exponential phase (~1x10 6 cells/mL), diluted to 2x10 4 cells/mL with AKseawater.After a 48-hour incubation, cells were mixed in a thermomixer for two minutes at 800 rpm at room temperature to dislodge any cells attached to biofilm at the bottom of the 96-well plate.100 µL of cells were transferred to Poly-D-Lysine (Sigma # P6407) coated 384-well imaging plates (Perkin Elmer Cell Carrier Ultra Plates #6057302) using an electronic multichannel pipetter (Rainin E4 Multi Pipette Multi E12-200XLS+).To optimize the screen, after 40 minutes of adherence, 1 µL of FM 1-43X mixture (1 µL of 500 µg/mL mixture of FM1-43X dye made by dissolving tube in 200 µL of methanol) was added to the cells and incubated for 15 minutes.50 µL of the cell dye-mixture was removed followed by fixation and washing as described next for the full screen.For the full screen, after 40 minutes of adherence, 50 µL of cells were removed, and the remaining 50 µL were washed once with 50 µL 4X PBS.After incubating in the 4X PBS for 5 minutes, 50 µL was removed.Cells were fixed for 20 minutes at room temperature by adding 50 µL of 4% formaldehyde in PEM buffer (100 mM PIPES pH 7, 1 mM EGTA, 1 mM MgSO4).After fixation, cells were washed by removing 50 µL of solution from the plate and adding 50 µL of PEM buffer three times.After the final wash, 75 µL of solution was removed.At this point, cells for the optimization screen were imaged on a Perkin Elmer Opera Phenix with the following imaging specifications for the Fluorescein (FITC) channel: 20X water objective (NA 1.0, working distance 1.7mm, 646 µm 2 field of view), 60ms and a three plane zstack at -8µm, -6µm and -4µm.For the full screen, after washing the fixative, cells were blocked by adding 75 µL of 2% BSA and 0.6% Triton-X100 in PEM for 30 minutes at room temperature.After blocking 25 µL was removed and 25 µL of primary antibody solution in 1% BSA and 0.3% Triton-X100 in PEM was added to stain the cell body and flagella overnight at 4°C (anti-tubulin, Abcam #ab6161, 1:1000 dilution).Due to the amount of time required for imaging, plates were staggered and processed one plate each day.On each imaging day, a plate was brought to room temperature and the primary antibody was washed three times by adding 50 µL of 1% BSA and 0.3% Triton-X100 in PEM and removing 50 µL of solution from the plate three times.Secondary antibody (Goat anti-rat Alexa Fluor 488, Invitrogen #A-11006, 1:300 dilution) and nuclear stain (DRAQ5, Thermo Scientific #62251, 1:500 dilution) were added in 25 µL of 1% BSA and 0.3% Triton-X100 in PEM and incubated for 2 hours at room temperature.After incubation, the secondary antibody was washed three times by adding 50 µL of PEM and removing 50 µL of solution from the plate three times.To stain the cell collar and obtain a second cell body marker, rhodamine phalloidin (Invitrogen #R415, 1:300 dilution) and Cell Tracker cmDIL (Invitrogen # C7000, 250 nM) were added for 25 minutes.To preserve the staining, 50 µL of solution was removed from the plate and 50 µL of 50% glycerol in PEM was added.22 fields of view in 4 planes (each plane separated by 1 µm) of each well in the 384-well plate were imaged with a 40X Water immersion objective (NA 1.1, working distance 0.62mm, 323 µm 2 field of view) on a Perkin Elmer Opera Phenix with optimized imaging specifications for each channel (Alexa 488 -Tubulin -20ms; TRITC -Cell Tracker cmDIL / rhodamine phalloidin -100ms; Brightfield -100ms; Alexa Fluor 647 -DRAQ5 / Nuclei -1s).Due to restrictions on the available image area on the opera phenix, columns 1-23 of each 384-well plate were imaged first followed by a second scan with column 24 alone.
For both assays, the quantified cell counts were normalized to the average cell count of all DMSO wells within an individual plate to account for any plate-to-plate variation.Compounds were determined to significantly inhibit S. rosetta cell proliferation if the resulting normalized cell count had a p-value < 0.05 (based on two-tailed p-value calculated from z-score of all treated samples).The resulting normalized cell count data were plotted using GraphPad Prism 9.3.1™(GraphPad San Diego, California, USA) (Figure 1; Figure S2B-C, Figure S2E, Figure S2G).
For 96-well plates, 1 µL of 37% formaldehyde was added to each well by using a multichannel pipette, and a pierceable aluminum plate seal (USA Scientific TempPlate ® Sealing Foil) was added to cover the plate.The plate was vortexed at 2000rpm (24-well plates) or 3000rpm (96well plates) in a plate vortexer (Eppendorf ThermoMixer C) to ensure equal fixation of cells in the well.Fixed cells were immediately counted or placed at 4°C for up to 2 weeks before counting.The cell density of sample timepoints along the full growth course was determined by analysis of micrographs taken on a Widefield microscope (Carl Zeiss AG Axio Observer.Z1/7, Oberkochen, Germany) 13 , or brightfield imaging using a cell counter (Logos Biosystems LUNA-FL™).The resulting normalized cell density data at each timepoint was plotted using GraphPad Prism 9.3.1™(GraphPad San Diego, California, USA) (Figure 2A; Figure 4A; Figure 4C; Figure S6B; Figure S10C).For comparisons between growth curves and phosphotyrosine signal (see Assessment of S. rosetta kinase signaling by western blotting) the area under the growth curve (AUC) was analyzed using GraphPad Prism 9.3.1™(GraphPad San Diego, California, USA) with baseline at Y=0 and minimum peak height > 10% above the baseline to maximum Y value (Figure S8; Figure S10B).
For dose-response assays, SrEpac cultured in high-nutrient media (5% Sea Water Complete 64 or 4% Sea Water Complete with 4% Cereal Grass 13 ) in exponential phase (~5x10 5 -9x10 5 cells/mL) was diluted to lower density.Starting density used varied based on treatment length: for 24-hour treatments and less, cells were plated at 2x10 5 cells/mL; for 24-48 hour treatments, cells were plated at 1x10 5 cells/mL; and for 48+ hour treatments, cells were plated at 5x10 4 cells/mL.Cell density was determined at the treatment endpoint using the same approach as the treatment growth curves described in the preceding paragraph.The resulting normalized cell density data at each dose was plotted using GraphPad Prism 9.3.1™(GraphPad San Diego, California, USA).(Figure 1B; Figure 4D; Figure S10D)

Assessment of S. rosetta kinase signaling by western blotting
After compound treatment, the SrEpac culture was harvested and lysed to quantify protein abundance and immunoblotting by western blot.S. rosetta cells were harvested by centrifugation at 6,000g for five minutes in Falcon tubes in a swinging bucket centrifuge and transferred into 1.5 mL Eppendorf tubes and washed two times with 4X phosphate buffered saline (PBS, 6.2 mM potassium phosphate monobasic, 621 mM sodium chloride, 10.8 mM sodium phosphate dibasic) with centrifugation at 6,000g for 5 minutes in a fixed angle centrifuge at room temperature in between each wash.Cells were resuspended and lysed in digitonin lysis buffer (20 mM Tris pH 8, 150 mM potassium chloride, 5 mM magnesium chloride, 250 mM sucrose, 1 mM Pefablock® SC serine protease inhibitor (Sigma-Aldrich Cat# 76307), 8 mM digitonin, 1 mM dithiothreitol, 0.06 U/µL benzonase nuclease, 1X Roche PhosSTOP phosphatase inhibitor cocktail, 1X Roche cOmplete protease inhibitor cocktail) for 30 minutes.

ActivX Mass Spectrometry workflow to identify S. rosetta proteins that bind sorafenib
For all mass spectrometry experiments, SrEpac cultures were grown to a high density in Pyrex baffled flasks without shaking.In tall necked flasks, cultures were grown in the maximum volume of culture media, and an aquarium pump was used to bubble air into the foam-plugged PYREX® Delong flask pierced with a serological pipette at a bubbling rate of approximately one bubble/second 66 .For wide 2.8L PYREX® Fernbach flasks, bubbling was not needed.S. rosetta cells were harvested by spinning in 200mL Nalgene conicals at 2000g for 10 minutes in a swinging bucket centrifuge at room temperature to obtain cell pellets.Cells were first washed by resuspending cell pellets from four conicals in approximately 45 mL of 4X PBS, transferring cells to 50 mL Falcon tubes, and spinning at 2000g for 10 minutes in a swinging bucket centrifuge at room temperature.Cells were washed a second time by resuspending cells in ~15 mL of 4X PBS per 50 mL Falcon tube, transferring cells to 15 mL Falcon tubes, and spinning at 2000g for 10 minutes in a swinging bucket centrifuge at room temperature.Cells were then ready for the ActivX probe workflow.
For ActivX probe enrichment, cells were lysed and 500 µL of 5 mg/mL supernatants were obtained as previously described 41 .20 mM of manganese chloride cofactor was added to the lysate and incubated for 5 minutes followed by the addition of 100 µM sorafenib or 1% DMSO (vehicle control) and incubation for 10 minutes.20 µM ActivX probe was added for kinase capture.Biotinylated proteins were captured using streptavidin beads in the presence of 6M urea / immunoprecipitation (IP) lysis buffer, and samples were washed with 6M urea / IP lysis buffer.Protein samples were provided to the University of California, Davis mass spectrometry facility (https://cmsf.ucdavis.edu)on beads and underwent standard tryptic digestion with Promega ProteaseMAX™ Surfactant, Trypsin Enhancer.Briefly, samples were first reduced at 56ºC for 45 minutes in 5.

Generation of Sr-p38 T110M strains by genome editing
Candidate guide RNA sequences were obtained for Sr-p38 using the EuPaGDT tool (http://grna.ctegd.uga.edu/) and the S. rosetta genome 19 hosted on Ensembl Protists (Ensembl 108) 68 .Guide RNA length was set at 15 and an NGG PAM sequence was used.Guide RNA candidates were filtered for guides with one on-target hit (including making sure the guides do not span exon-exon boundaries), zero off-target hits (including against the genome of the cocultured bacterium E. pacifica), lowest strength of the predicted secondary structure (assessed using the RNAfold web server: http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi), and annealing near codon 110 of Sr-p38.A crRNA with the guide sequence CTACATCATCACAGAGAAGA, as well as universal tracrRNAs, were ordered from IDT (Integrated DNA Technologies, Coralville, IA).Repair templates were designed as single-stranded DNA oligos, in the same sense strand as the guide RNA, with 50 base pairs of genomic sequence on either side of the DSB cut site.The repair oligo GGTTGACCTGTACATCTCCGACGCGCGTGACATCTACATCATCATGGAGAAGATGGTGTGT ATCTTTGACGCGGGTTGACTGGCCGTGATGGCGCGTGTT was ordered from IDT as an Ultramer.Genome editing proceeded as described previously (Coyle 2023).

Bioinformatic analysis of the S. rosetta kinome
The S. rosetta kinome was annotated based on previously predicted kinases 19 and orthoDB ortholog annotation 69 of S. rosetta and human protein sequences in Uniprot 70 (Figure S1).For prediction of choanoflagellate p38 kinases, a HMMER profile was generated from human and previously predicted S. rosetta p38 kinases 19,45 and searched against available genomes and transcriptomes of choanoflagellates 9,12,71,72 (Figure S6A-B).Protein targets of individual human kinase inhibitors were manually annotated and plotted using CORAL 73 (Figure S1).References for kinase inhibitory data for each compound is available in Table S1.To analyze conservation within the kinase domain, kinase sequence alignments of predicted kinases were generated with Clustal Omega 74 (Figure S5B-C; Figure S9B) and amino acid logos were generated with WebLogo 75 (Figure S5A).S1 for the list of compounds, cell counts, and annotated human kinase targets) resulted in a distribution of cell counts, assessed by flow cytometry, at the 24-hour endpoint.S. rosetta cell counts were normalized to DMSO controls (dark grey).Compounds determined to significantly inhibit S. rosetta cell proliferation (based on two-tailed p-value < 0.05 calculated from z-score) are indicated in red, compounds that were not detected as significant inhibitors by flow cytometry but were identified by imaging (based on two-tailed p-value < 0.05 calculated from zscore) are in blue and compounds that were not significant inhibitors for either screen are indicated in light grey.Z-score statistics of the assays are indicated in Figure S2B    The ActivX ATP probe was used to pull down kinases from S. rosetta lysates that were pretreated with either DMSO or the ATP-competitive inhibitor sorafenib.We found that pretreatment with sorafenib reduced the level of p38 recovered using the ActivX ATP probe, indicating that sorafenib and p38 interact and outcompete ActivX ATP probe binding.Kinases plotted are only those that were identified in 855 both vehicle and sorafenib pre-treatments.For full kinase enrichment list, see Table S2, and for alignment of S. rosetta p38 with those from animals and fungi, see Fig. S9.
5 mM DTT followed by alkylation for one hour in the dark with iodoacetamide added to a final concentration of 10 mM.Trypsin was added at a final enzyme:substrate mass ratio of 1:50 and digestion carried out overnight at 37ºC.The reaction was quenched by flash freezing in liquid nitrogen and the digest was lyophilized.Digest was reconstituted in 0.1% TFA with 10% acetonitrile prior to injection.For quantification of peptides within each sample, 500 fmol of Hi3 E. coli standard reagent ClpB (Waters, Milford, MA) was placed into each sample and injected with 1.0 ug of total digest.Each sample was run in triplicate.The mass spectrometry instrument used to analyze the samples was a Xevo G2 QTof coupled to a nanoAcquity UPLC system (Waters, Milford, MA).Samples were loaded onto a C18 Waters Trizaic nanotile of 85 µm × 100 mm; 1.7 μm (Waters, Milford, MA).The column temperature was set to 45°C with a flow rate of 0.45 mL/min.The mobile phase consisted of A (water containing 0.1% formic acid) and B (acetonitrile containing 0.1% formic acid).A linear gradient elution program was used: 0-40 min, 3-40 % (B); 40-42 min, 40-85 % (B); 42-46 min, 85 % (B); 46-48 min, 85-3 % (B); 48-60 min, 3% (B).Mass spectrometry data were recorded for 60 minutes for each run and controlled by MassLynx 4.2 SCN990 (Waters, Milford, MA).Acquisition mode was set to positive polarity under resolution mode.Mass range was set form 50 -2000 Da.Capillary voltage was 3.5 kV, sampling cone at 25 V, and extraction cone at 2.5 V. Source temperature was held at 110C.Cone gas was set to 25 L/h, nano flow gas at 0.10 Bar, and desolvation gas at 1200 L/h.Leucine-enkephalin at 720 pmol/µl (Waters, Milford, MA) was used as the lock mass ion at m/z 556.2771 and introduced at 1 µL/min at 45 second intervals with a 3 scan average and mass window of +/-0.5 Da.The MSe data were acquired using two scan functions corresponding to low energy for function 1 and high energy for function 2. Function 1 had collision energy at 6 V and function 2 had a collision energy ramp of 18 − 42 V. RAW MSe files were processed using Protein Lynx Global Server (PLGS) version 3.0.3(Waters, Milford, MA).Processing parameters consisted of a low energy threshold set at 200.0 counts, an elevated energy threshold set at 25.0 counts, and an intensity threshold set at 1500 counts.Each sample was searched against the Salpingoeca rosetta genome hosted on Ensembl Genomes 19,67 .Each databank was randomized within PLGS and included the protein sequence for ClpB.Possible structure modifications included for consideration were methionine oxidation, asparagine deamidation, glutamine deamidation, serine dehydration, threonine dehydration, and carbamidomethylation of cysteine.For viewing, PLGS search results were exported in Scaffold v4.4.6 (Proteome Software Inc., Portland, OR).

Figure 1 .
Figure 1.High-throughput screening of a small molecule library revealed inhibitors of S. rosetta cell proliferation.(A) Treatment of S. rosetta cultures with 1255 different small molecules (see TableS1for the list of compounds, cell counts, and annotated human kinase targets) resulted in a distribution of cell counts, assessed by flow cytometry, at the 24-hour endpoint.S. rosetta cell counts were normalized to DMSO controls (dark grey).Compounds determined to significantly inhibit S. rosetta cell proliferation (based on two-tailed p-value < 0.05 calculated from z-score) are indicated in red, compounds that were not detected as significant inhibitors by flow cytometry but were identified by imaging (based on two-tailed p-value < 0.05 calculated from zscore) are in blue and compounds that were not significant inhibitors for either screen are indicated in light grey.Z-score statistics of the assays are indicated in FigureS2Band Figure S2E.Sorafenib (SO), the focus of this study, is labeled.(B) The range of normalized cell counts for compounds that significantly inhibited S. rosetta cell proliferation by flow cytometry.Compounds that were the focus of further study -genistein (GE), glesatinib (GL), PP121, masitinib (MA), sotrastaurin (SOT) -are labeled.Z-score statistics of the assays are indicated in Figure S2B.(C) Comparison of normalized values of compounds that inhibited S. rosetta cell proliferation, assessed by flow cytometry and their corresponding normalized values by imaging.Compounds that were the focus of further study -sorafenib (SO), genistein (GE), glesatinib (GL), PP121, masitinib (MA) -are labeled.
Figure 1.High-throughput screening of a small molecule library revealed inhibitors of S. rosetta cell proliferation.(A) Treatment of S. rosetta cultures with 1255 different small molecules (see TableS1for the list of compounds, cell counts, and annotated human kinase targets) resulted in a distribution of cell counts, assessed by flow cytometry, at the 24-hour endpoint.S. rosetta cell counts were normalized to DMSO controls (dark grey).Compounds determined to significantly inhibit S. rosetta cell proliferation (based on two-tailed p-value < 0.05 calculated from z-score) are indicated in red, compounds that were not detected as significant inhibitors by flow cytometry but were identified by imaging (based on two-tailed p-value < 0.05 calculated from zscore) are in blue and compounds that were not significant inhibitors for either screen are indicated in light grey.Z-score statistics of the assays are indicated in FigureS2Band Figure S2E.Sorafenib (SO), the focus of this study, is labeled.(B) The range of normalized cell counts for compounds that significantly inhibited S. rosetta cell proliferation by flow cytometry.Compounds that were the focus of further study -genistein (GE), glesatinib (GL), PP121, masitinib (MA), sotrastaurin (SOT) -are labeled.Z-score statistics of the assays are indicated in Figure S2B.(C) Comparison of normalized values of compounds that inhibited S. rosetta cell proliferation, assessed by flow cytometry and their corresponding normalized values by imaging.Compounds that were the focus of further study -sorafenib (SO), genistein (GE), glesatinib (GL), PP121, masitinib (MA) -are labeled.

Figure 2 .
Figure 2. Glesatinib and sorafenib, two multi-target tyrosine kinase inhibitors, disrupt S. rosetta cell proliferation and tyrosine phosphosignaling.(A) Treatment of S. rosetta cultures with 1 µM sorafenib and glesatinib led to a complete block of cell proliferation, while treatment with 1 µM masitinib or PP121 led to a partial reduction in cell proliferation relative to DMSO-treated cultures.Two biological replicates were conducted per treatment, and each point represents the mean of three measurements from each biological replicate.For timepoints at 40, 60, and 85 hours, cell densities of inhibitor-treated cultures were significantly different from vehicle (DMSO) (p-value <0.01).Significance was determined by a two-way ANOVA multiple comparisons test.(B) S. rosetta cultures treated with 1 µM or 10 µM sorafenib, glesatinib, or PP121 for 24 hours had reduced normalized cell density, whereas masitinib only had reduced normalized cell density at 10 µM.Normalized cell densities were determined to be reduced if differences between treatments and vehicle (DMSO) were significant (p-value <0.01) Significance was determined by determined by a two-way ANOVA multiple comparisons test.Movies show S. rosetta cells treated with 10 µM glesatinib that undergo cell lysis (Movie S1) and sorafenib, that have cell body deformation (Movies S2-S3), in comparison to DMSO control (Movie S4).(C) Western blot analysis of S. rosetta cultures treated with 1 µM sorafenib and glesatinib for 1 hour showed a decrease in tyrosine phosphorylation of proteins at ~60kDa, ~45kDa, and ~35kDa (indicated by arrows and detected with pY1000 anti-phosphotyrosine antibody) compared to vehicle (DMSO) control.Masitinib and PP121 did not reduce the phosphotyrosine signal.

Figure 4 .
Figure 4. Inhibitors of human p38 kinases disrupt S. rosetta cell proliferation through multiple kinase targets.(A) Human p38 kinase inhibitors disrupt S. rosetta cell proliferation.S. rosetta cultures were treated with sorafenib or one of two human p38-specific inhibitors, skepinone-L or BIRB 796, in 24-well plates over an 80-hour growth course.At 40 hours, cells treated with 10 µM skepinone-L, BIRB 796 or sorafenib showed little evidence of cell proliferation in comparison to vehicle (DMSO) control (p-value <0.01).Cells treated with 1 µM skepinone-L or BIRB 796 had reduced cell density in comparison to vehicle (DMSO) control (pvalue <0.01) at 60 hours.Three biological replicates were conducted per experiment and significance was determined by determined by a two-way ANOVA multiple comparisons test.(B)Strategy for generating Sr-p38 T110M cell lines.The S. rosetta p38 locus was targeted by a guide RNA complexed with Cas9 that anneals near the "gatekeeper" amino acid codon 110 in the kinase domain and directs Cas9 to introduce a double-strand break (DSB) downstream of codon 110.The Cas9-guide RNA complex was coupled with a homology-directed repair template to insert a cassette that encodes a threonine to methionine mutation at position 110.In humans, this mutation desensitizes p38 to sorafenib, and skepinone-L 44 .(C) Two independent clones maintained with cycloheximide (CHX) resistance, CHX Sr-p38 T110M-1 , and CHX Sr-p38 T110M-2 , do not show a growth defect when compared CHX 1 and CHX 2 , two strains that are only cycloheximide-resistant. Three technical replicates were conducted per strain, and significance was determined by a two-way ANOVA multiple comparisons test.(D) CHX Sr-p38 T110M strains do not show decreased sensitivity to human p38 kinase inhibitors.Two independent CHX Sr-p38 T110M and CHX strains were treated with sorafenib, skepinone-L, and BIRB 796, and cells were counted after 24 hours.Normalized counts did not differ between the