Selection of library compoundsAll compounds for the propofluor arm library (PFAL) were designed around a fluorobenzene core, with variable alkyl ‘arms’. One could generate a massive library solely with modifications to alkyl carbon number, position, and shape, but the final 13-member PFAL (Fig. 1) consisted of a subset of non-chiral alkyl arms (methyl, ethyl, isopropyl and tertbutyl) in three different positions (1, 1,3 and 1,4) on the fluorobenzene ring to create a library with systematic alterations in size and configuration of the alkyl carbon side chains. The proposed compound 12 (1,3-ditertbutyl-2-fluorobenzene) was unable to be synthesized and thus not included in the library. Physiochemical parameters for these compounds can be found in Table S1.Figure 1Chemical Structures of the Propofluor Arm Library (PFAL) Compounds and Commonly Used Anesthetics. This library of compounds is composed of alkyl derivatives of the molecule propofluor (compound 9, previously called fropofol). The intended compound 12 was unable to be synthesized and thus not characterized. Chemical names and abbreviations are as follows: 1: fluorobenzene (FB), 2: 1-methyl-2-fluorobenzene (1-Me-2-FB), 3: 1,3-dimethyl-2-fluorobenzene (1,3-diMe-2-FB), 4: 1,4-dimethyl-2-fluorobenzene (1,4-diMe-2-FB), 5: 1-ethyl-2-fluorobenzene (1-Et-2-FB), 6: 1,3-diethyl-2-fluorobenzene (1,3-diEt-2-FB), 7: 1,4-diethyl-2-fluorobenzene (1,4-diEt-2-FB), 8: 1-isopropyl-2-fluorobenzene (1-iPr-2-FB), 9: 1,3-diisopropyl-2-fluorobenzene (1,3-diiPr-FB) 10: 1,4-diisopropyl-2-fluorobenzene (1,4-diiPr-FB), 11: 1-tertbutyl-2-fluorobenzene (1-ditBu-FB), 12: 1,3-ditertbutyl-2-fluorobenzene (1,3-ditBu-FB), 13: 1,4-ditertbutyl-2-fluorobenzene (1,4-ditBu-FB). The chemical structures of 3 commonly used anesthetics (Propofol, Sevoflurane and Isoflurane) are shown in the lower left corner.Toxicity of PFAL compoundsAn important initial step in the testing these compounds is an assessment of the toxicity. The small size of 5 dpf (days post fertilization) larval zebrafish (< 4 mm total body length) facilitates bulk diffusion for gas transport, thus toxicity due to respiratory depression is inaccessible and therefore a limitation of this model compared to adult fish and mammals that rely on breathing and bulk flow respiration19,20. Nevertheless, an LD50 (median lethal dose) in this model remains a useful means to compare toxicity, and also prevents mistakenly characterizing mortality as the decreased movement observed with anesthetic states. In this assay, PFAL compounds were found to be considerably less toxic than propofol itself (Fig. S1), including some with LD50 values that were unable to be determined due to both low toxicity and limited solubility. For these compounds (1–5) an LD50 was estimated with available experimental data and an upper constraint of 100% mortality. All concentrations of PFAL compounds used in later testing were far below these LD50 values. Co-administration 30 μM PFAL compounds with propofol yielded a minimal increase (< 6 μM changes) in LD50 that remains far above propofol’s clinically useful concentrations.Activity of the PFAL compoundsEach compound was tested for sedative activity when administered alone prior to combination testing with propofol (Fig. 2a-d). PFAL compounds alone at 10 and 30 μM revealed no statistically significant decrease in spontaneous movement (SM) or elicited movement (EM) that would indicate sedation or anesthesia. Although no decrease in SM or EM was observed for any PFAL compound, this does not rule out sub-clinical sedation that might be observed with propofol co-administration. On the other hand, compounds 7, 9, and 10 demonstrated increased SM, and only compounds 9 and 10 showed increased EM. To further characterize this increase in SM/EM, we compared the pattern of activity to that produced by pentylenetetrazole21, a known epileptogenic compound, and the observed increase in SM was not consistent with seizure. Pentylenetetrazole administration results in a period of normal movement, followed by a short period of uncontrolled movement (seizure) and then decreased movement characteristic of a post-ictal state. This is in contrast to the PFAL treated fish where this dose-dependent increase in movement was simply more rapid movement (Fig. S2 and Supplemental Video). A change in movement due to chemical irritation is another consideration, but previous data using mustard oil shows that chemical irritation actually results in decreased movement17. Ultimately, the mechanism of the observed increase in movement is not clear but may relate to activation of specific motor networks.Figure 2Screening of PFAL compounds for antagonism of anesthesia. PFAL compounds were screened in 5 dpf zebrafish for their ability to antagonize the anesthetic effects of propofol (panels a, b, c and d) as well as sevoflurane and isoflurane (panels e and f). Each dot represents the data collected from one fish. Spontaneous movement (SM) is the distance of spontaneous swimming in a 4 min period, and elicited movement (EM) is distance moved in 1 s after exposure to a tap stimulus. Both measures were scaled to same-day DMSO controls. Concentrations of drugs used for screening were based on previously determined EC50 values: Propofol SM = 0.01 μM, EM= 2.8 μM; Sevoflurane SM = 76 μM, EM = 240 μM; Isoflurane SM = 42, μM EM = 180 μM17. Mean and 95% CI are shown for each data set. Statistical comparison with both DMSO (D) only (black asterisks) and anesthetic control (Ctrl, gray asterisks) groups are shown. Note that each DMSO comparison is to the DMSO only control (sample on the furthest left of each graph, and the anesthetic control comparisons are made within each anesthetic grouping as indicated by the bars along the x-axis of each graph. ns: P > 0.05, *: P ≤ 0.05, **: P ≤ 0.01, ***: P ≤ 0.001, ****: P ≤ 0.0001).Antagonism of propofol anesthesiaBased on previously reported EC50 values of propofol in 5 dpf zebrafish (propofol EC50 for SM = 0.1 μM, EC50 for EM = 3.8 μM)17, propofol concentrations of 0.3, 1, and 5 μM were co-administered with PFAL compounds (up to 30 μM) as an initial test of anesthetic additivity or antagonism. Compounds 7, 9 and 10 administered at 30 μM showed antagonism of propofol via measures of SM, but only compounds 9 and 10 antagonized propofol at a concentration of 5 μM (Fig. 2a). To further test the potency of these compounds, the PFAL concentration was tested at 10 μM and with this lower dose only 10 demonstrated antagonism of propofol (Fig. 2b). Despite having the most antagonistic potency for SM, even 10 did not show antagonism at a propofol concentration of 5 μM. However, at a concentration of 10 μM, 10 was able to antagonize propofol at 1 μM which is about 10-times higher than propofol’s SM EC50 of 0.1 μM. This initial screen seems to suggest compound 10 is more potent as a propofol antagonist compared to the previously characterized compound 9 (propofluor).The pattern of SM antagonism demonstrated by the PFAL compounds suggests that there may be a ‘sweet spot’ of hydrophobicity and/or sterics that makes an important contribution to antagonistic potency. It is well described that relative LogP of anesthetics correlates to potency, if hydrophobicity alone were a primary drive of potency, one would expect to see a bell-shaped trend with a peak indicating ideal configuration and/or hydrophobicity. However, an analysis of the estimated EC50 derived from this initial screening (see Fig. S3) compared to the calculated LogP shows little variation amongst the antagonism of PFAL compounds, with the obvious exceptions of 9 and 10 (see Fig. 3). This pattern suggests that the geometry of the carbon chains has a greater contribution to antagonist potency than simple hydrophobicity. Interestingly, although much less potent than 10 by measures of SM, 7 was the third most potent antagonizing compound, and it shares the same ‘1,4’ alkyl chain configuration as 10. This may additionally suggest differences compared to the ‘1,3’ alkyl configuration. However, it is also notable that when examined for antagonism via measures of EM (Fig. 2c-d), 7 was the only compound that did not antagonize propofol at 30 μM PFAL concentration (Fig. 2c). Overall, there is poor correlation in this initial screen between antagonism of propofol by SM and by EM. At 30 μM concentrations for EM, all the PFAL compounds (except 7) showed statistically significant antagonism of propofol. Even at 10 μM PFAL, only the EM of 2 and 11 were found not to be statistically different than the propofol controls. The inability to distinguish large differences in potency for EM based on structural differences supports mechanistic differences in antagonism, and by proxy differences in the underlying anesthetic mechanism of these movements. However, even with compounds that showed antagonism, this initial screen by a measure of EM (response to tap stimulus) was not found to correspond to a statistically significant rightward shift in the EC50 of propofol for compounds 9 and 10 (see Fig. 4). Thus, there are obvious limitations in what conclusions with respect to anesthetic antagonism can be draw from the single concentration tested in the screen, but it does not negate the interest of these findings as there are multiple important anesthetic endpoints and the antagonism of propofol via measure of SM is worth further consideration.Figure 3Comparison of LogP and Estimated Antagonistic Potency. The EC50 of propofol in the presence of PFAL compounds (as indicated in the legend) were estimated from the screening data in Fig. 2. These values were compared to calculated LogP values for each compound. Linear regressions (95% CI shown) exclude the data from propofol only (P). The gray regression excludes the compounds that demonstrated statistically significant antagonism via SM (7, 9, and 10) and the blue regression excludes the other compounds that did not demonstrate antagonism.Figure 4Propofol EC50 curves with PFAL 9 and 10. a Spontaneous movement of propofol alone and with the addition of 30 μM 9 or 10. Propofol alone had an EC50 of 0.14 μM (95% CI 0.10–0.20), and both 9 and 10 showed statistically significant rightward shifts in EC50. 9: 1.4 μM (95% CI 1.10–1.88), P < 0.0001 (~ tenfold increase); 10: 0.58 μM (95% CI 0.35–0.75), P < 0.0001 (~ fourfold increase). b Elicited movement of propofol alone and with the addition of 30 μM 9 or 10. Propofol alone had an EC50 of 3.53 μM (95% CI 2.29–5.13), neither 9 nor 10 showed statistically significant shifts in EC50. 9: 4.38 μM (95% CI 3.22–6.06), P = 0.50; 10: 5.00 μM (95% CI 2.85–5.91), P = 0.23.Antagonism of sevoflurane and isoflurane anesthesiaTo assess the selectivity of this anesthetic antagonism, a smaller group of PFAL compounds was tested for antagonism of two volatile anesthetics sevoflurane and isoflurane. As with propofol, these anesthetic concentrations were initially chosen for each drug based on previously reported EC50 values (Sevoflurane SM = 76 μM, EM = 240 μM; Isoflurane SM = 42 μM, EM = 180 μM)17, and PFAL doses of 30 μM. No evidence of antagonism of either volatile anesthetic was found (see Fig. 2e-f) via measures of SM or EM. These findings suggest some selectivity for antagonism of propofol anesthesia, and render unlikely the idea that antagonism is arising from non-specific stimulation. This is similar to previous findings that indicate the opposite; some stimulants having greater potency in antagonizing volatile anesthetics compared to propofol8, lending further support for underlying differences in mechanisms of action for these anesthetic classes.Antagonism of anesthesia after inductionIn the above studies, the anesthetic and PFAL compounds were co-administered to the fish simultaneously. In order to show that these compounds can antagonize ‘anesthesia’ once established, a sequential addition was conducted whereby the fish were exposed to propofol (3 μM) for 10 min, the solution was removed and replaced with solution containing propofol (3 μM) and either compound 9 or 10 (30 μM). Although this experiment does not provide precise kinetic data, the exposure to either 9 or 10 produced rapid reversal of propofol immobility despite the continued presence of an immobilizing concentration of propofol (Fig. 5a-c).Figure 5Further Characterization of Propofol Antagonism with Compounds 9 and 10. a and b 5 dpf zebrafish (n = 36) were first anesthetized with 3 μM propofol and then 30 μM of either compound 9 or 10 was added. Between minutes 5 and 10 there is a small gap in the data during which the propofol solution was removed and a solution containing propofol and PFAL compound was added. Spontaneous movement of the fish resumed very shortly after compound addition. c 2% DMSO vehicle control for panels A and B. n = 36 for panels A-C (each dot represents 1 fish). d IC50 of compounds 9 and 10 measured in 5 dpf zebrafish exposed to 3 μM Propofol was found to be 37.5 μM and 25.3 μM respectively. The mean and 95% CI is shown for each data point in all panels.IC50 of compounds 9 and 10. The initial screening experiments revealed 9 and 10 as the most potent anesthetic antagonists in the PFAL library. At a propofol concentration of 3 μM, IC50 (median inhibitory concentration) values were determined to quantitatively compare their potency (Fig. 5d). The IC50 values were closer than the initial screening experiments suggested (IC50 of 9 = 37.5 μM, 10 = 25.3 μM); however, the finding is less surprising when one considers the striking difference in Hill slopes (9 = 1.8, 10 = 65), may also be an additional clue to that these compounds have mechanistic differences.Inhibition of 1-AMA binding to a model proteinThe ability of anesthetics, to inhibit 1-aminoanthracene (1-AMA) binding to the ‘anesthetic site’ of model protein HSAF (horse spleen apoferritin) has been shown to correlate to anesthetic activity22,23,24. However, it is known that 9 also binds to this site14, suggesting the assay may also reveal optimal physicochemistry of other antagonists. Each compound in the library inhibited the binding of 1-AMA to HSAF (indicated by decreased fluorescence from unbound 1-AMA), except for compound 13 where testing was limited by solubility (Fig. 6). Calculation of IC50 values would suggest that compounds 9 and 10 are the most potent inhibitors of 1-AMA binding. However, the shape of these IC50 curves for 9 and 10 are markedly different from the other PFAL library members in that they demonstrate incomplete inhibition of 1-AMA binding which is similar to the binding seen by propofol in this system (Fig. S4). This similarity to propofol is consistent with the hypothesis of competitive antagonism of propofol. The appearance of incomplete competition at soluble concentrations is not explained by artifact of 1-AMA/PFAL interactions (Fig. S5) and is not readily explained by our current understanding of anesthetic binding in this system. Despite this, there is a similarity between the interactions of propofol, 9, and 10 that differentiates the interactions of these compounds from the other PFAL analogues.Figure 6Binding of PFAL Compounds to Apoferritin. The relative binding affinity of PFAL compounds to HSAF (horse spleen apoferritin) was measured by displacement of the fluorophore 1-AMA (1-aminoanthracene) that demonstrates increased fluorescence when bound to HSAF. Error bars represent 95% CI of the calculated IC50, but upper limits were unable to be estimated for compounds 1, 3, 4, 6, or 7. nb = non-binding (a) Calculated IC50 of PFAL compounds. (b) Binding curves from which IC50 values were calculated. Mean and standard deviation for each data point are shown.Schild analysisTo begin an in vivo investigation of potential mechanisms of antagonism, a Schild analysis was conducted for both compounds 9 and 10. EC50 curves for propofol with the addition of 0, 20, 30, 40, and 50 μM 9 and 10 were obtained (Fig. 7a-b), and then used to form the basis of Schild Plots (Fig. 7c-d). Compounds 9 and 10 exhibit overall linear correlations which is suggestive of competitive binding in a Schild analysis25; however, such a conclusion from this analysis would require all assumptions of a Schild analysis be met26, which is not the case in this system. A Schild analysis requires that an antagonist acts only at a single receptor type, but general anesthetics have many mechanistically relevant targets. Additionally, in a Schild plot for a competitive antagonist, a linear correlation of antagonism with a parallel rightward shift is seen across multiple orders of magnitude. This is in contrast to the plots found here for 9 and 10 that are linear across merely tens of micromolar concentrations (Fig. 7c-d) and do not show parallel rightward shifts (Fig. 7a-b) and thus cannot be said to be classical competitive antagonists. The slope of the linear fit is also > 1 for both compounds, which is often attributed to non-specific binding, which in some cases can be attributed to adsorption to glassware, but given the known non-specific binding of propofol in complex biological systems it is not surprising that these molecules could exhibit similar behavior to non-specific adsorption. Overall, this analysis must be viewed with caution as even if compounds 9 and 10 did act at a single site, propofol most likely does not which could explain the observation that these antagonists do not exhibit the “parallel shift” seen for a classical competitive antagonist. It is also important to note that these findings are observations of SM. A parallel analysis was conducted with observations of the mechanistically distinct EM, but no such relationship of antagonism was observed even with application of higher concentrations of 9 and 10 (see Fig. S6).Figure 7Schild Analysis of Propofol Antagonism with Compounds 9 and 10. (a,b) EC50 curves of propofol (SM) with co-administration of compounds 9 and 10 at 0, 20, 30, 40, and 50 μM. Error bars indicate 95% CI. (c,d) Schild plots derived from the EC50 data in sub-panels (a,b).Co-administration of compounds 9 and 10To further examine the potential differences between compounds 9 and 10, an additivity study was performed with both compounds co-administered with propofol at 15 μM each (for a total PFAL concentration of 30 μM). Compared to either compound administered alone there is an approximately threefold increase in EC50 indicating a synergism between these two compounds (Fig. 8). Given the complex and non-classic pharmacology of anesthetics, it is difficult to propose a specific mechanism of these antagonists from such an observation. Like anesthetics, these antagonists likely bind at multiple sites and multiple targets, but this synergism and the other data presented here support that these two lead compounds do not have identical mechanisms of action.Figure 8Synergism of compounds 9 and 10. To test for additivity, the antagonism of propofol by compounds 9 and 10 alone (30 μM) were compared to a combined dose of 9 and 10 (15 μM each) with a total PFAL concentration of 30 μM. Compared to 9 alone (EC50 = 1.3 μM), the combination of antagonists (9 + 10) at an equivalent concentration showed a threefold increase in antagonism (EC50 = 4.0 μM). Error bars indicate 95% CI.
