Sacharomyces cerevisiae strain HSP104-GFP (BY4741HSP104-GFP-HIS3-MX6) (Invitrogen) cells were grown in Yeast Nitrogen Base (YNB, 6.7 g/l) with Complete Supplement Mixture (CSM, 1.54 g/l) and 2% glucose (pH = 6), at 30˚C on a shaker (220 rpm). The cells were collected at OD₆₀₀ = 0.6 - 1.0.

The selective Hog1 inhibitor 4-(1-benzyl-4-phenyl-1H-1,2,3-triazol-5-yl)-N-isopropylpyridin-2-amine (M = 478.67 g/mol) [29] [31] was dissolved in dimethyl sulfoxide (DMSO) to a final concentration of 1.9 mM stock solution. The final inhibitor concentration in YNB was 10 µm and incubation time was set to 5 minutes.

Sodium arsenite (NaAsO₂) (Sigma-Aldrich) was dissolved in water prior to use and applied to a final concentration of 0.1 mM in YNB.

The microfluidic device was fabricated using soft lithography as previously described [31] [43] . Just prior the experiment, glass syringes (Hamilton Company) containing the cells and different media respectively were attached to the four-inlet microfluidic flow chamber via polytetrafluoro ethylene tubings (Cole-Parmer, Vernon Hills, IL, USA). Using a fiber-coupled ytterbium laser (YLD-5-LP, IPG Photonics) with a wavelength of 1070 nm and estimated power of 240 mW in the laser focus [39] , single cells were trapped and moved from the flow of the cell inlet channel. One by one, cells were then gently pressed against the bottom of the system to adhere and form a 5 × 5 cell array. The trapping procedure limited the cells exposure to the infrared light only for a few seconds (<10 s) to ensure minimized photo damage [44] [45] . The location of the cell array (120 µm, −10 µm) and the flow rates were determined by simulations of the velocity profiles and the concentration gradients in the system based on the diffusion coefficients of the fluids (Figures 1(a) and (b)) to ensure that cells are not subjected to any substrate gradients but are fully covered by the specific treatment. Microenvironmental changes necessary for monitoring the Hsp104-GFP redistribution (aggregate formation), were created by changing the flow rates in the inlet channels (5, 5, 500 and 1000 nL/min for channels “1”, “2”, “3” and “4”, respectively) set by the syringe pumps (CMA 400, CMA Microanalysis). In this way, the environment could rapidly be changed from neutral media to a media containing Hog1 inhibitor, and thereafter to a media containing arsenite. The customwritten automation software in Open Lab (PerkinElmer, Waltham, MA, USA) used for trapping and positioning the

cells as well as controlling the syringe pumps, the microscope and the EM-CCD camera (C9100-12, Hamamatsu Photonics) used for the image acquisition, enabled an automated setup and maximized the control of the experimental process. For determining the number of aggregates, cells were imaged as z-stacks with nine focal layers spaced 0.8 µm apart. These images were acquired before and every 60 seconds for 5 minutes during inhibitor treatment or before and every 5 minutes for 60 minutes during arsenite treatment. The images were analyzed using CellStress software [46] and statistics were done using non-parametric Kruskal-Wallis tests.

To examine whether treating cells with Hog1 kinase inhibitor results in increased arsenite influx into the cells, we compared the arsenite induced aggregate formation as a readout for increase in arsenite influx into the cells [33] in control and inhibitor treated cells. To address this issue, we first established that the condition in microfluidic device in the presence of only growth media does not cause any aggregate formation (Figure 2). We next asked whether the Hog1 kinase inhibitor per se triggers Hsp104-GFP redistribution (aggregate formation). Based on previous study inhibiting Hog1 kinase activity with the applied inhibitor [31] , (Hamngren, C. et al., submitted), a range of concentrations between 5 μM to 25 μM of Hog1 kinase inhibitor and different incubation times were examined, of which the final concentration of 10 μM for a period of 5 minutes incubation was chosen as increasing concentration/incubation time could result in false aggregate formation. We monitored Hsp104-GFP relocalization in the cells positioned in the array upon exposure to growth media for 45 minutes after 5 minutes pretreatment with 10 μM inhibitor. Our data clearly showed that the applied inhibitor concentration and period of incubation does not promote aggregate formation within the investigated condition and time span (Figure 3(a)). Although the total period of the experiments were initially set to 60 minutes, as some cells showed slight increase in aggregate formation around 50 minutes, the time lapse studies were instead performed to terminate at 45 minutes to avoid any false aggregate formation. The late aggregate formation following incubation with Hog1 inhibitor might be explained by different scenarios. There is a possibility that incubation with the inhibitor

either directly target the components involved in protein folding or result in stress conditions [32] e.g. oxidative stress [47] that triggers Hsp104 relocalization. The Hog1 inhibitor may also be partly non-specific and hence target other kinases in the cell resulting in eventual protein misfolding and Hsp104 redistribution. However, addressing the specificity of kinase inhibitors in yeast is rather difficult as panels of purified kinases are not commercially available for such screens, whereas available for human kinases [48] . Another likely explanation would be that inhibition of the kinase activity of the Hog1 for a longer time leads to protein misfolding and thereby induction of heat shock protein responses. Hog1 kinase activity has been shown required to provide a necessary function to cope with unfolded protein accumulation due to ER stress [49] . Considering the fact that Hog1 becomes activated upon a wide variety of environmental cues and is involved in transcription, translation, transport, as well as cell cycle adaptations in response to different conditions [23] [24] [29] , these are all plausible scenarios. These mechanisms are remained to be explored further in the future studies. However, in order to avoid conceivable synergetic aggregate formation, we have excluded the later time points when the few inhibitor induced aggregates occur.

We next asked if arsenite is efficiently taken up by the cells when introduced in the flow and if it induces quantifiable Hsp104-GFP redistribution (aggregates) as reported before [33] in our setup. To this end, the cell arrays were subjected to the final concentration of 0.1 mM sodium arsenite and the aggregate formation was followed by time lapse imaging every 5 minutes for 45 minutes. The results (Figure 3(b)) showed that arsenite is effectively taken up by cells (46 cells), induces Hsp104-GFP relocalization and arsenite induced aggregates were precisely quantifiable with the CellStress software [46] . The arsenite induced aggregate formation initiated roughly at 30 minutes after stress exposure. The maximum number of aggregates per cell found was six, the fraction of the cells showing aggregates after 45 minutes was slightly below 50% ranging from 1 to 4 aggregates per cell (Figure 3(b)) and the mean value of aggregates per cell was 0.85 (data not shown). As it was described earlier [33] we correlated the number of induced aggregates to the uptake/intracellular level of arsenite. This assumption is bolstered by the fact that the induced aggregates indeed increase by arsenite concentration (Figure 2, [33] ) whereas disappear upon over expression of arsenite export pump [33] . Our data also demonstrating that aggregate formation is not induced in normal condition in the presence of growth media (Figure 2) supports that the aggregate formation seen in arsenite-rich condition is likely not due to other stresses such as heat or cold stress but to the fact that the cells reside in arsenite supplemented growth media. However, this assertion could be supplemented and strengthened by traditional arsenite uptake assays. Nonetheless, the arsenite interference with protein folding and consequent aggregate formation does not contradict previously identified arsenite induction of oxidative stress [50] and could also be the case in our experiments.

To address whether the Hog1 kinase inhibitor is able to improve the arsenite uptake, we compared the arsenite induced aggregate formation in terms of increase in the number of cells having aggregates, increase in the number of aggregates per cell and time of initiation of aggregate formation in inhibitor treated and untreated conditions. Cell arrays were exposed to 10 µm Hog1 inhibitor for 5 minutes and subsequently to 0.1 mM sodium arsenite for 45 minutes. Images were acquired prior and then every minute during the 5 minutes inhibitor treatment. Subsequently, images were attained every 5 minutes during the arsenite exposure through the end of the time lapse experiment. Interestingly, the arsenite induced aggregate formation following pretreatment with the Hog1 kinase inhibitor initiated earlier (20 minutes compared to 30 minutes). The maximum number of aggregate per cell in this condition reached up to nine (compared to six), the fraction of the cells displaying aggregate formation was about 80% after 45 minutes (compared to 50%) (Figure 3(c)) and the mean value of aggregates per cell (of in total 86 cells) was 2.35 times higher (2 vs. 0.85) (data not shown).

Hence, we concluded that 1) the Hog1 inhibitor does not induce aggregate formation per se in the investigated time span and condition. However, as discussed earlier, the few aggregates formed at later time points might be due to inhibitor interference with protein folding, inhibitor induced stress conditions, the possibility of the inhibitor being non-specific that in the long run could affect protein folding components or a consequence of longterm Hog1 inhibition; 2) the arsenite induced Hsp104-GFP redistribution [33] is reproducible when arsenite is introduced in the flow and is quantifiable in our setup. As previous and current data confirm, the aggregate formation is indeed induced by arsenite, however, the mechanisms of Hsp104-GFP induction upon arsenite exposure has been described elsewhere [33] , and most importantly; 3) the short-term Hog1 kinase inhibitor treatment of the cells prior to arsenite treatment results in faster and increased aggregate formation both in terms of number of cells showing aggregate as well as number of aggregates per cell, probably due to the augmented arsenite uptake.

We are the first to provide an approach in which small molecule, cell-permeable, fast acting and highly efficient kinase inhibitor can act as powerful means to significantly (Figure 3(d)) elevate the arsenite induced aggregates supposedly through elevated arsenite influx into the cells without any requirement for introducing genetic modifications. Although, mechanisms behind this effect is not exclusively elucidated and requires further investigations, the current results together with our previous study [23] favors a possible mechanism (Figure 4) in which Fps1is negatively regulated by a kinase activate Hog1 [23] . However, in the presence of the inhibitor the kinase activity of Hog1 is partially impaired, which in turn leads to higher arsenite uptake via Fps1 indicated by higher number of arsenite induced aggregates. This hypothesis could be further evidenced e.g. by applying a truncated version of Fps1 that lacks the N-terminal regulatory domain rendering Fps1 constitutively open [51] . However, other players i.e. the possibility of inhibitor induced aggregate formation or other stress conditions such as oxidative stress induction of Hsp104-GFP relocalization are not ruled out (Figure 4). The fact that increase in arsenite uptake and thereby increase in the number of aggregates is partial is perhaps partly due to the incomplete inhibition of Hog1 [31] and partly due to the other existing arsenite detoxification mechanisms e.g. arsenite export pump Acr3 [52] and/or arsenite-glutathione conjugation and sequestration to the vacuole through the ABC transporter Ycf1 [53] (Figure 4) in the cell. Regardless of the cause, the partial nature of this MAPK inhibition might have advantages as it prevents massive arsenite influx and severe, irreversible damage to the cells.

The cell-to-cell variation even exists in genetically homogenous populations. These variations may arise from intrinsic (differences in the expression levels of genes in the same cell) or extrinsic (cell-to-cell variation in a population) sources [54] [55] . The latter can partly be explained by fluctuations that occur through different cell cycle stages [56] . Therefore, we asked if the variation in response between mother cells and buds can to some extent explain the cell-to-cell variation in our results. To this end, we discriminated between mother and daughter cells (only attached buds) and examined whether the As(III) induced Hsp104-GFP relocalization (aggregate) shows a bimodal distribution between mother cells and their buds. The results showed no induced aggregate formation in mother cells (54 cells) or buds (28 cells) following inhibitor treatment (Figure 5(a)). Upon arsenite treatment, the number of aggregates was not substantially (Figure 5(d)) different in the mother cells (33 cells) compared to the buds (13 cells) (40% compared to the 60% respectively, Figure 5(b)), (mean value of 0.72 compared to 1.1 per cell respectively, data not shown). Whereas, this difference was slightly more pronounced (Figure 5(d)) following kinase inhibitor treatment (mean value of 2.5 vs. 1.2 respectively, data not shown). The fraction of the mother cells (55 cells) showing aggregates compared to the buds (31 cells) was 90% to 60% respectively and the maximum number of aggregates per cell was also higher in mother cells compared to the buds (9 compared to 5) (Figure 5(c)). More interestingly, the effect of kinase inhibitor treatment in increasing arsenite uptake was much more significant (Figure 5(d)) when it was studied only on discriminated mothers compared to indiscriminated cells (mothers and buds together). While the mean number of aggregates in each mother cell upon exposure to only arsenite was 0.7 the mean number of aggregates per each mother cell upon arsenite

treatment post inhibitor exposure was 2.5 (data not shown) proving 3.6 times increase. This significant difference (Figure 5(d)) is very well reflected in the fraction of the mother cells showing aggregates upon arsenite treatment (40% Figure 5(b)) compared to the mother cells exposed to arsenite following inhibitor treatment (90% Figure 5(c)). In addition, the maximum number of aggregates following arsenite treatment after 45 minutes reaches up to 4 (Figure 5(b)) whereas upon inhibitor-arsenite exposure the maximum number of aggregates reaches up to 9 aggregates (Figure 5(c)). Collectively, our data indicates that the arsenite induced aggregates post kinase inhibitor treatment are more accumulated in mother cells compared to the buds. This bimodal behavior may simply be a result of lower abundance of Fps1 in the plasma membrane of buds, different Hog1 to Fps1 protein expression ratios in buds compared to mother cells or higher uptake efficiency of inhibitors by mothers compared to buds. It can also be in line with the earlier observation where asymmetric inheritance of oxidatively damaged proteins and their accumulation in mother cells was reported to guarantee the fitness of newly born cells [57] . Previously, it has been shown that the otherwise cytosolic evenly distributed Hsp104-GFP redistributes to distinct foci, proved to be sites of protein aggregates [33] -[35] . Arsenite interferes with protein folding, induces protein aggregates and Hsp104 co-sediments with arsenite induced protein aggregates [33] . This complex here called as Hsp104-GFP aggregates seems to be asymmetrically distributed between mother and daughter cells, which might also be a way to provide fitness to the buds.