The Parallel Structure–Activity Relationship Screening of Three Compounds Identifies the Common Agonist Pharmacophore of Pyrrolidine Bis-Cyclic Guanidine Melanocortin-3 Receptor (MC3R) Small-Molecule Ligands

The melanocortin receptors are involved in numerous physiological pathways, including appetite, skin and hair pigmentation, and steroidogenesis. In particular, the melanocortin-3 receptor (MC3R) is involved in fat storage, food intake, and energy homeostasis. Small-molecule ligands developed for the MC3R may serve as therapeutic lead compounds for treating disease states of energy disequilibrium. Herein, three previously reported pyrrolidine bis-cyclic guanidine compounds with five sites for molecular diversity (R1–R5) were subjected to parallel structure–activity relationship studies to identify the common pharmacophore of this scaffold series required for full agonism at the MC3R. The R2, R3, and R5 positions were required for full MC3R efficacy, while truncation of either the R1 or R4 positions in all three compounds resulted in full MC3R agonists. Two additional fragments, featuring molecular weights below 300 Da, were also identified that possessed full agonist efficacy and micromolar potencies at the mMC5R. These SAR experiments may be useful in generating new small-molecule ligands and chemical probes for the melanocortin receptors to help elucidate their roles in vivo and as therapeutic lead compounds.


Introduction
The melanocortin family consists of five known receptor subtypes (numbered 1-5) [1-8], endogenous agonists including α-MSH, β-MSH, γ-MSH, and ACTH derived from the proopiomelanocortin gene transcript [9], and the naturally occurring antagonists agouti and agouti-related protein (AGRP) [10][11][12][13][14][15]. This receptor family is involved in many biological functions that are linked to specific melanocortin receptor subtypes, such as pigmentation with the MC1R [2,6], steroidogenesis with the MC2R [6], and exocrine gland function in rodents with the MC5R [16]. The MC4R has been extensively studied for its role in food intake and appetite. Individuals with MC4R loss-of-function polymorphisms present with an obese phenotype [17,18], a phenotype that is also observed in MC4R KO mice [19]. The MC3R is similarly linked to energy homeostasis. Mice that lack the MC3R have an increased fat mass-to-lean mass ratio while maintaining a similar body weight to wild-type littermates [20,21]. Reduced food intake has also been reported for male MC3R KO mice maintained on a regular chow diet, with increased feed efficiency observed for Herein, three previously identified pyrrolidine bis-cyclic guanidine compounds were selected as leads for initial SAR studies [29]. For each of the starting compounds, a series of truncated and stereochemical analogs were made to explore the importance of the five substitution positions (R1-R5). All compounds were screened for agonist activity at the MC1R, MC3R, MC4R, and MC5R. Since all compounds with full MC3R efficacy did not possess full MC4R efficacy, these ligands were also assayed for antagonist activity at the MC4R. Synthesizing and screening the compound series in parallel allowed for the identification of the common pharmacophore unit correlating with full agonist efficacy at the MC3R, and to examine if SAR trends in one ligand were present in the other compounds.

Results
Three compounds (2718.001, 2718.002, and 2718.003; Figure 1) were selected from a library of previously reported pyrrolidine bis-cyclic guanidine compounds (corresponding to the prior publication compounds numbered 11, 2, and 23, respectively [29]). Whereas all of the previously reported compounds possessed a Pro to generate the pyrrolidine functionality [29], the stereochemistry of this group was varied in the present study, introducing a new substitution point. While the R1 functionality remains constant between the prior publication and the experiments reported herein, the R2 position herein represents the stereochemistry of the pyrrolidine group. Therefore, the prior publication groups R2, R3, and R4 are reported herein as R3, R4, and R5, respectively. Compounds 2718.001 and 2718.002 both possess R1 R-isobutyl, R2 S-pyrrolidine, R4 (S,S)-1-hydroxyethyl, R5 4-tbutyl-cyclohexyl-methyl functionalities, and vary at the R3 position (R-benzyl in 2718.001 and R-cyclohexyl-methyl in 2718.002; Figure 1). Both compounds were reported to possess similar mMC3R agonist potency (EC50 = 310 nM for 2718.001 and 210 nM for 2718.002) [29]. Performing parallel SAR studies on both ligands probes the importance of the R3 position for activity, as well as if the R3 position alters the minimal pharmacophore necessary for full MC3R agonist activity. Altering the R1 position from R-isobutyl to R-isopropyl while maintaining the same R2, R3, R4, and R5 groups as 2718.001 and 2718.002 identified additional compounds with similar mMC3R potencies [29]. Whereas compounds with an R1 R-isobutyl group required an R4 (S,S)-1-hydroxyethyl for observable mMC3R agonist potency, several compounds with an R1 R-isopropyl group and an R4 R-propyl or R-isopropyl possessed full agonist efficacy at the mMC3R [29]. The most potent of these compounds (2718.003 herein) possessed an R1 R-isopropyl, Scheme 1.

Results
Three compounds (2718.001, 2718.002, and 2718.003; Figure 1) were selected from a library of previously reported pyrrolidine bis-cyclic guanidine compounds (corresponding to the prior publication compounds numbered 11, 2, and 23, respectively [29]). Whereas all of the previously reported compounds possessed a Pro to generate the pyrrolidine functionality [29], the stereochemistry of this group was varied in the present study, introducing a new substitution point. While the R1 functionality remains constant between the prior publication and the experiments reported herein, the R2 position herein represents the stereochemistry of the pyrrolidine group. Therefore, the prior publication groups R2, R3, and R4 are reported herein as R3, R4, and R5, respectively. Compounds 2718.001 and 2718.002 both possess R1 R-isobutyl, R2 S-pyrrolidine, R4 (S,S)-1-hydroxyethyl, R5 4-tbutylcyclohexyl-methyl functionalities, and vary at the R3 position (R-benzyl in 2718.001 and R-cyclohexyl-methyl in 2718.002; Figure 1). Both compounds were reported to possess similar mMC3R agonist potency (EC 50 = 310 nM for 2718.001 and 210 nM for 2718.002) [29]. Performing parallel SAR studies on both ligands probes the importance of the R3 position for activity, as well as if the R3 position alters the minimal pharmacophore necessary for full MC3R agonist activity. Altering the R1 position from R-isobutyl to R-isopropyl while maintaining the same R2, R3, R4, and R5 groups as 2718.001 and 2718.002 identified additional compounds with similar mMC3R potencies [29]. Whereas compounds with an R1 R-isobutyl group required an R4 (S,S)-1-hydroxyethyl for observable mMC3R agonist potency, several compounds with an R1 R-isopropyl group and an R4 R-propyl or R-isopropyl possessed full agonist efficacy at the mMC3R [29]. The most potent of these compounds (2718.003 herein) possessed an R1 R-isopropyl, R2 S-pyrrolidine, R3 R-cyclohexyl-methyl, R4 R-propyl, and R5 4-tbutyl-cyclohexyl-methyl (mMC3R EC 50 = 350 nM) [29]. Due to the incorporated R1 and R4 changes, this compound was selected as another starting compound to determine the importance of these substitutions for MC3R activity (Figure 1). R2 S-pyrrolidine, R3 R-cyclohexyl-methyl, R4 R-propyl, and R5 4-tbutyl-cyclohexyl-methyl (mMC3R EC50 = 350 nM) [29]. Due to the incorporated R1 and R4 changes, this compound was selected as another starting compound to determine the importance of these substitutions for MC3R activity (Figure 1). Several parallel modifications to the three compounds were performed to identify SAR trends. Functional groups were serially truncated off each compound to identify the importance of each substitution point, as well as generating R4 and R5 fragments for each scaffold. The stereochemistry at individual positions was also sequentially varied to identify any potential effects of altering chirality. The enantiomer form of each starting compound (where all stereochemical centers were inverted) was also generated to examine the effects of globally altering the stereochemistry. Specifically for 2718.001, analogs consisting of two and three positions with inverted stereocenters were generated. Due to the reported similar mMC3R potencies of 2718.001 and 2718.002, varying only at the R3 position (the aromatic R-benzyl and aliphatic R-cyclohexyl-methyl groups, respectively), a set of compounds was generated that varied aromatic (R-4-hydroxy-benzyl, R-(1H-indol-3yl)-methyl, and R-2-methyl-naphthalene) and aliphatic (R-methyl, R-butyl, and R-cyclohexyl) substitutions at the R3 position while maintaining R1 R-isobutyl, R2 S-pyrrolidine, R4 (S,S)-1-hydroxyethyl, and R5 4-tbutyl-cyclohexyl-methyl functionalities.
Individual pyrrolidine bis-cyclic guanidine compounds were synthesized on a solid support using N-α-Boc compatible chemistry (Scheme 1), as previously described [29]. Traditional solid-phase peptide synthesis generated capped peptides in parallel. While on the solid support, the backbone amides were transformed into the corresponding amines by an established borane reduction of the carbonyl without epimerization of the sidechain α-carbon [29,33,34]. Following the reduction step, cyclization of the polyamine functionalities with cyanogen bromide yielded the desired pyrrolidine bis-cyclic guanidine compounds. The final products were globally sidechain deprotected and cleaved off the polystyrene resin support with anhydrous hydrogen fluoride (HF). After removal of the excess HF, the individual compounds were purified using HPLC. Of the 52 synthesized compounds, 50 were purified to >95% purity, as determined by peak integration at 214 nM. One compound (2718.002) was purified to >94% purity, and one compound (2718.024) was purified to >84% purity. The compounds were characterized by 1 H NMR and LC/MS (Supplemental Information).
Pyrrolidine bis-cyclic guanidine compounds were assessed for biological activity using the AlphaScreen cAMP kit with HEK293 cells stably expressing the mMC1R, mMC3R, mMC4R, or mMC5R, as previously described [37][38][39]. Since the MC2R is only stimulated by ACTH and no other endogenously derived POMC agonists [40], it was not examined in the current study. The synthetic melanocortin agonist peptide NDP-MSH was included Several parallel modifications to the three compounds were performed to identify SAR trends. Functional groups were serially truncated off each compound to identify the importance of each substitution point, as well as generating R4 and R5 fragments for each scaffold. The stereochemistry at individual positions was also sequentially varied to identify any potential effects of altering chirality. The enantiomer form of each starting compound (where all stereochemical centers were inverted) was also generated to examine the effects of globally altering the stereochemistry. Specifically for 2718.001, analogs consisting of two and three positions with inverted stereocenters were generated. Due to the reported similar mMC3R potencies of 2718.001 and 2718.002, varying only at the R3 position (the aromatic R-benzyl and aliphatic R-cyclohexyl-methyl groups, respectively), a set of compounds was generated that varied aromatic (R-4-hydroxy-benzyl, R-(1H-indol-3-yl)-methyl, and R-2methyl-naphthalene) and aliphatic (R-methyl, R-butyl, and R-cyclohexyl) substitutions at the R3 position while maintaining R1 R-isobutyl, R2 S-pyrrolidine, R4 (S,S)-1-hydroxyethyl, and R5 4-tbutyl-cyclohexyl-methyl functionalities.
Individual pyrrolidine bis-cyclic guanidine compounds were synthesized on a solid support using N-α-Boc compatible chemistry (Scheme 1), as previously described [29]. Traditional solid-phase peptide synthesis generated capped peptides in parallel. While on the solid support, the backbone amides were transformed into the corresponding amines by an established borane reduction of the carbonyl without epimerization of the sidechain αcarbon [29,33,34]. Following the reduction step, cyclization of the polyamine functionalities with cyanogen bromide yielded the desired pyrrolidine bis-cyclic guanidine compounds. The final products were globally sidechain deprotected and cleaved off the polystyrene resin support with anhydrous hydrogen fluoride (HF). After removal of the excess HF, the individual compounds were purified using HPLC. Of the 52 synthesized compounds, 50 were purified to >95% purity, as determined by peak integration at 214 nM. One compound (2718.002) was purified to >94% purity, and one compound (2718.024) was purified to >84% purity. The compounds were characterized by 1 H NMR and LC/MS (Supplemental Information).
Pyrrolidine bis-cyclic guanidine compounds were assessed for biological activity using the AlphaScreen cAMP kit with HEK293 cells stably expressing the mMC1R, mMC3R, mMC4R, or mMC5R, as previously described [37][38][39]. Since the MC2R is only stimulated by ACTH and no other endogenously derived POMC agonists [40], it was not examined in the current study. The synthetic melanocortin agonist peptide NDP-MSH was included as a control compound. Compounds were assayed for agonist activity in a 7-point doseresponse curve (10 −3 to 10 −9 M). Ligands that possessed full mMC3R agonist efficacy (relative to NDP-MSH) and partially activated the mMC4R were assayed for antagonist activity using the synthetic melanocortin agonist NDP-MSH in a Schild experimental paradigm [41]. Pyrrolidine bis-cyclic guanidines that did not possess agonist activity in two independent experiments were considered inactive at the concentrations assayed and were not further studied. Active compounds were assayed in at least three independent experiments. Compounds that possessed sigmoidal dose-response curves and partial agonist efficacy (<90% NDP-MSH signal) were binned into two groups (A = 20-50% NDP-MSH, B = 51-90% NDP-MSH; Figure 2). Due to the inherent experimental error associated with these types of assays, we considered compounds that were at within a 3-fold potency range to be equipotent. Since the AlphaScreen cAMP assay is a loss-of-signal assay, in which higher concentrations of compound result in lower assay signal, the data were normalized to baseline and maximal NDP-MSH signal for illustrative purposes, as previously described [38,39]. as a control compound. Compounds were assayed for agonist activity in a 7-point doseresponse curve (10 −3 to 10 −9 M). Ligands that possessed full mMC3R agonist efficacy (relative to NDP-MSH) and partially activated the mMC4R were assayed for antagonist activity using the synthetic melanocortin agonist NDP-MSH in a Schild experimental paradigm [41]. Pyrrolidine bis-cyclic guanidines that did not possess agonist activity in two independent experiments were considered inactive at the concentrations assayed and were not further studied. Active compounds were assayed in at least three independent experiments. Compounds that possessed sigmoidal dose-response curves and partial agonist efficacy (<90% NDP-MSH signal) were binned into two groups (A = 20-50% NDP-MSH, B = 51-90% NDP-MSH; Figure 2). Due to the inherent experimental error associated with these types of assays, we considered compounds that were at within a 3-fold potency range to be equipotent. Since the AlphaScreen cAMP assay is a loss-of-signal assay, in which higher concentrations of compound result in lower assay signal, the data were normalized to baseline and maximal NDP-MSH signal for illustrative purposes, as previously described [38,39].  Tables 1-4. "A" indicates a partial agonist response resulting in 20-50% receptor activation, and "B" indicates a partial agonist response resulting in 50-90% receptor activation. A "Full Agonist" response indicates that the compound resulted in >90% receptor activation. All percentages are relative to the NDP-MSH agonist as a 100% reference compound.

Structure-Activity Relationship Studies of 2718.001
The first lead compound (2718.001) was previously reported to possess full mMC3R and mMC5R agonist activity (EC50 = 310 and 530 nM, respectively), partially stimulated the mMC1R (58% efficacy at 10 µM concentrations) and the mMC4R (10-50% activation at 10 µM concentrations), and was an antagonist at the mMC4R (pA2 = 7.0 with agonist MTII) [29] when assayed in a β-galactosidase reporter gene assay [42]. In the present experiment (assayed with the cAMP AlphaScreen kit), full agonist efficacy was observed at the mMC3R and mMC5R (EC50 = 270 and 260 nM, respectively; Table 1 and Figure 3). Full agonist activity was also observed for the mMC1R (EC50 = 320 nM; Figure 3), unlike the partial response previously observed. This difference may be due to the different assays used to examine activity. Partial activation of the mMC4R was observed in the present study (60% at 1 mM concentrations), and 2718.001 possessed antagonist activity at the mMC4R (pA2 = 6.0 with agonist NDP-MSH; Figure 4), supporting the MC3R agonist/MC4R partial response and antagonist response of 2718.001 previously reported.  Tables 1-4. "A" indicates a partial agonist response resulting in 20-50% receptor activation, and "B" indicates a partial agonist response resulting in 50-90% receptor activation. A "Full Agonist" response indicates that the compound resulted in >90% receptor activation. All percentages are relative to the NDP-MSH agonist as a 100% reference compound.

Structure-Activity Relationship Studies of 2718.001
The first lead compound (2718.001) was previously reported to possess full mMC3R and mMC5R agonist activity (EC 50 = 310 and 530 nM, respectively), partially stimulated the mMC1R (58% efficacy at 10 µM concentrations) and the mMC4R (10-50% activation at 10 µM concentrations), and was an antagonist at the mMC4R (pA 2 = 7.0 with agonist MTII) [29] when assayed in a β-galactosidase reporter gene assay [42]. In the present experiment (assayed with the cAMP AlphaScreen kit), full agonist efficacy was observed at the mMC3R and mMC5R (EC 50 = 270 and 260 nM, respectively; Table 1 and Figure 3). Full agonist activity was also observed for the mMC1R (EC 50 = 320 nM; Figure 3), unlike the partial response previously observed. This difference may be due to the different assays used to examine activity. Partial activation of the mMC4R was observed in the present study (60% at 1 mM concentrations), and 2718.001 possessed antagonist activity at the mMC4R (pA 2 = 6.0 with agonist NDP-MSH; Figure 4), supporting the MC3R agonist/MC4R partial response and antagonist response of 2718.001 previously reported.

Double Residue Stereocenter Inversion Analogs
R1: R-isobutyl R2: S-pyrrolidine R3: S-benzyl R4: (S,S)-1-hydroxyethyl R5: 4-tbutyl-cyclohexyl-methyl 2300 ± 900 55% @ 1 mM 55% @ 1 mM 70% @ 1 mM 75% @ 1 mM 50% @ 1 mM 55% @ 1 mM 3000 ± 1000 a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC50 or pA2 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two independent experiments. For partial agonist compounds (a sigmoidal dose-response curve was observed, but efficacy was less than 90% maximal NDP-MSH signal), compounds were binned into two groups: A (20-50% receptor activation) or B (51-90% receptor activation), Figure 2. For compounds possessing full agonist efficacy at the MC3R and that did not fully activate the MC4R, follow-up antagonist experiments and corresponding pA2 values, via a Schild analysis [41], were performed at the mMC4R. R3: S-benzyl R4: (R,R)-1-hydroxyethyl R5: 4-tbutyl-cyclohexyl-methyl 75% @ 1 mM 50% @ 1 mM 55% @ 1 mM 3000 ± 1000 a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC50 or pA2 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two independent experiments. For partial agonist compounds (a sigmoidal dose-response curve was observed, but efficacy was less than 90% maximal NDP-MSH signal), compounds were binned into two groups: A (20-50% receptor activation) or B (51-90% receptor activation), Figure 2. For compounds possessing full agonist efficacy at the MC3R and that did not fully activate the MC4R, follow-up antagonist experiments and corresponding pA2 values, via a Schild analysis [41], were performed at the mMC4R.
R1: S-isobutyl R2: R-pyrrolidine R3: S-benzyl R4: (R,R)-1-hydroxyethyl R5: 4-tbutyl-cyclohexyl-methyl 75% @ 1 mM 50% @ 1 mM 55% @ 1 mM 3000 ± 1000 a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC 50 or pA 2 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC 50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two independent experiments. For partial agonist compounds (a sigmoidal dose-response curve was observed, but efficacy was less than 90% maximal NDP-MSH signal), compounds were binned into two groups: A (20-50% receptor activation) or B (51-90% receptor activation), Figure 2. For compounds possessing full agonist efficacy at the MC3R and that did not fully activate the MC4R, follow-up antagonist experiments and corresponding pA 2 values, via a Schild analysis [41], were performed at the mMC4R. Replacing the R4 (S,S)-1-hydroxyethyl with hydrogen (2718.022) resulted in equipotent agonist activity at the mMC1R (EC50 = 400 nM), mMC3R (EC50 = 380 nM), and mMC5R (EC50 = 320 nM) compared to 2718.001, with similar partial activation (50% at 1 mM concentrations) and antagonist activity (pA2 = 6.1) at the mMC4R. Similar agonist potency at the mMC1R (EC50 = 390 nM) and mMC5R (EC50 = 480 nM) and mMC4R antagonist activity (pA2 = 6.3) were also observed when replacing the R1 R-isobutyl group with hydrogen (2718.024) compared to 2718.001, although an 11-fold decrease in mMC3R potency (EC50 = 3000 nM) was also observed for this substitution. Truncation of the R3 position to hydrogen (2718.023) or the R5 position to ethyl (2718.004) did not result in compounds with full agonist efficacy, although 2718.023 induced a partial agonist response at the mMC4R (EC50 = 40,000 nM) and 2718.004 did result in partial agonism at the mMC1R (EC50 = 20,600; Figure 3). Replacing both the R4 and R5 positions with hydrogen (2718.006) resulted in minimal-to-no activity at 1 mM concentrations at all of the receptors, while incorporation Replacing the R4 (S,S)-1-hydroxyethyl with hydrogen (2718.022) resulted in equipotent agonist activity at the mMC1R (EC50 = 400 nM), mMC3R (EC50 = 380 nM), and mMC5R (EC50 = 320 nM) compared to 2718.001, with similar partial activation (50% at 1 mM concentrations) and antagonist activity (pA2 = 6.1) at the mMC4R. Similar agonist potency at the mMC1R (EC50 = 390 nM) and mMC5R (EC50 = 480 nM) and mMC4R antagonist activity (pA2 = 6.3) were also observed when replacing the R1 R-isobutyl group with hydrogen (2718.024) compared to 2718.001, although an 11-fold decrease in mMC3R potency (EC50 = 3000 nM) was also observed for this substitution. Truncation of the R3 position to hydrogen (2718.023) or the R5 position to ethyl (2718.004) did not result in compounds with full agonist efficacy, although 2718.023 induced a partial agonist response at the mMC4R (EC50 = 40,000 nM) and 2718.004 did result in partial agonism at the mMC1R (EC50 = 20,600; Figure 3). Replacing both the R4 and R5 positions with hydrogen (2718.006) resulted in minimal-to-no activity at 1 mM concentrations at all of the receptors, while incorporation Replacing the R4 (S,S)-1-hydroxyethyl with hydrogen (2718.022) resulted in equipotent agonist activity at the mMC1R (EC 50 = 400 nM), mMC3R (EC 50 = 380 nM), and mMC5R (EC 50 = 320 nM) compared to 2718.001, with similar partial activation (50% at 1 mM concentrations) and antagonist activity (pA 2 = 6.1) at the mMC4R. Similar agonist potency at the mMC1R (EC 50 = 390 nM) and mMC5R (EC 50 = 480 nM) and mMC4R antagonist activity (pA 2 = 6.3) were also observed when replacing the R1 R-isobutyl group with hydrogen (2718.024) compared to 2718.001, although an 11-fold decrease in mMC3R potency (EC 50 = 3000 nM) was also observed for this substitution. Truncation of the R3 position to hydrogen (2718.023) or the R5 position to ethyl (2718.004) did not result in compounds with full agonist efficacy, although 2718.023 induced a partial agonist response at the mMC4R (EC 50 = 40,000 nM) and 2718.004 did result in partial agonism at the mMC1R (EC 50 = 20,600; Figure 3). Replacing both the R4 and R5 positions with hydrogen (2718.006) resulted in minimal-to-no activity at 1 mM concentrations at all of the receptors, while incorporation of the cyclic guanidine containing the R4 (S,S)-1-hydroxyethyl and R5 4-tbutyl-cyclohexylmethyl groups resulted in full mMC5R agonist activity (EC 50 = 140,000 nM; Figure 3).
Inverting single stereocenters of 2718.001 resulted in two compounds (2718.033 possessing an R1 S-isobutyl group and 2718.031 possessing an R2 R-pyrrolidine group) that possessed equipotent mMC1R and mMC5R agonist potencies relative to 2718.001. These two compounds also possessed partial agonist efficacies and micromolar potencies at the mMC3R, and partially activated the mMC4R. Inverting the R3 position to S-benzyl (2718.031) decreased mMC1R agonist potency 7-fold relative to 2718.001, and partially activated the mMC3R, mMC4R, and mMC5R. Changing the stereochemistry at the R4 position to (R,R)-1-hydroxyethyl resulted in a compound that was a partial agonist at the mMC1R (EC 50 = 4000 nM) and mMC5R (EC 50 = 5000 nM), while partially activating the mMC3R and mMC4R.
Simultaneously inverting the stereochemistry of two positions resulted in three equipotent mMC1R agonist compounds (2718.040 altering R1 and R2; 2719.038 altering R2 and R3, Figure 3; 2718.036 altering R2 and R4) relative to 2718.001. These double substitutions are the three double substitutions that invert the stereochemistry at the R2 position. Altering the R3 and R4 stereochemistry (2718.035) decreased mMC1R potency 12-fold relative to 2718.001, while partial agonist efficacy was observed for altering the R1 and R3 (2718.039) and R1 and R4 (2718.037) positions. None of the double stereochemical inversion compounds resulted in full agonist efficacy at the mMC3R or mMC4R. Two compounds (2718.038 and 2718.040) were full agonists at the mMC5R. The R1-and R2-inverted 2718.040 was an equipotent mMC5R agonist relative to 2718.001, while the 2718.038 (inverting the R2 and R3 positions) possessed 5-fold decreased potency at the mMC5R ( Figure 3) relative to 2718.001.

Structure-Activity Relationship Studies of 2718.002
The second compound examined, 2718.002, was previously reported to possess full mMC1R, mMC3R, and mMC5R agonist activity (EC 50 = 300, 210, and 140 nM, respectively), partial agonist efficacy at the mMC4R (45% MTII signal, EC 50 = 270 nM), and antagonist activity at the mMC4R (pA 2 = 5.8 with MTII agonist) [29]. The results are similar to the present report (Table 2) where full agonist activity was observed at the mMC1R (EC 50 = 270 nM), mMC3R (EC 50 = 240 nM), and mMC5R (EC 50 = 290 nM). A partial agonist response was observed at the mMC4R (EC 50 = 1200 nM), and 2718.002 possessed antagonist activity in the presence of NDP-MSH (pA 2 = 6.2 with NDP-MSH agonist). These data support the observed mMC3R agonist and mMC4R partial agonist/antagonist pharmacology previously reported.        (B) 65% @ 1 mM 65% @ 1 mM 2800 ± 100 a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC50 or pA2 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two R1: R-isobutyl R2: S-pyrrolidine R3: R-cyclohexyl-methyl R4: (R,R)-1-hydroxyethyl R5: 4-tbutyl-cyclohexyl-methyl 1000 ± 100 3500 ± 200 (B) 50% @ 1 mM 2200 ± 700  (B) 65% @ 1 mM 65% @ 1 mM 2800 ± 100 a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC50 or pA2 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two independent experiments. For partial agonist compounds (a sigmoidal dose-response curve was observed, but efficacy was less than 90% maximal NDP-MSH signal), compounds were binned into two groups: A (20-50% receptor activation) or B (51-90% receptor activation), Figure 2. For compounds possessing full agonist efficacy at the MC3R and that did not fully activate the MC4R, follow-up antagonist experiments and corresponding pA2 values, via a Schild analysis [41], were performed at the mMC4R.
R1: S-isobutyl R2: R-pyrrolidine R3: S-cyclohexyl-methyl R4: (R,R)-1-hydroxyethyl R5: 4-tbutyl-cyclohexyl-methyl 8000 ± 2000 (B) 65% @ 1 mM 65% @ 1 mM 2800 ± 100 a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC 50 or pA 2 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC 50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two independent experiments. For partial agonist compounds (a sigmoidal dose-response curve was observed, but efficacy was less than 90% maximal NDP-MSH signal), compounds were binned into two groups: A (20-50% receptor activation) or B (51-90% receptor activation), Figure 2. For compounds possessing full agonist efficacy at the MC3R and that did not fully activate the MC4R, follow-up antagonist experiments and corresponding pA 2 values, via a Schild analysis [41], were performed at the mMC4R.
Inverting the stereochemistry of the R1 position in this scaffold to S-isobutyl (2718.048) resulted in an equipotent mMC1R (EC 50 = 390 nM) and mMC5R (EC 50 = 650 nM) agonist compared to 2718.002. This compound was a partial agonist at the mMC3R (EC 50 = 2900 nM) and partially activated the mMC4R (65%) at 1 mM concentrations. Changing the R2 stereochemistry to R-pyrrolidine (2718.047) also maintained mMC1R and mMC5R potency relative to 2718.002. This compound did not fully stimulate the mMC3R and was a partial agonist at the mMC4R (EC 50 = 3100 nM). Changing the R3 position to S-cyclohexyl-methyl (2718.046) or the R4 position to R,R-1-hydroxyethyl (2718.045) resulted in decreased mMC1R (7-and 4-fold, respectively) and mMC5R (17-and 8-fold, respectively) agonist potency compared to 2718.002. Both compounds were partial agonists at the mMC3R, and partially activated the mMC4R at 1 mM concentrations. While this series did not examine double or triple stereochemical inversions, changing the stereochemistry at the R1, R2, R3, and R4 positions (2718.049) resulted in a partial agonist at the mMC1R (EC 50 = 8000 nM), partial activation of the mMC3R and mMC4R (65% at 1 mM concentrations), and was a full agonist at the mMC5R (EC 50 = 2800 nM), possessing 10-fold decreased potency relative to 2718.002.

Structure-Activity Relationship Studies at the R3 Position of 2718.001 and 2718.002
The 2718.001 and 2718.002 lead compounds both possess an R1 R-isobutyl, R2 Spyrrolidine, R4 (S,S)-1-hydroxylethyl, and R5 4-tbutyl-cyclohexyl-methyl groups. The 2718.001 bis-cyclic guanidine possesses an R3 R-benzyl, while 2718.002 has an R3 Rcyclohexyl-methyl functionality. A set of six substitutions (Table 3), comprised of aliphatic and aromatic substitutions, were incorporated into the R3 position. Both 2818.001 and 2718.002 possess similar mMC1R (EC 50 = 320 and 270 nM, respectively), mMC3R (EC 50 = 270 and 240 nM, respectively), and mMC5R (EC 50 = 260 and 290 nM, respectively) agonist potencies. The R3-substituted compounds are compared to 2718.002 herein, but can be compared to 2718.001 with similar relative activities.     a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC50 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two independent experiments. For partial agonist compounds (a sigmoidal dose-response curve was observed, but efficacy was less than 90% maximal NDP-MSH signal), compounds were binned into two groups: A (20-50% receptor activation) or B (51-90% receptor activation), Figure 2.
R1: R-isobutyl R2: S-pyrrolidine R3: R-cyclohexyl R4: (S,S)-1-hydroxyethyl R5: 4-tbutyl-cyclohexyl-methyl 70% @ 1 mM 11,000 ± 4000 (A) 55% @ 1 mM 6000 ± 1000 a Pyrrolidine bis-cyclic guanidine compounds were assayed at the selected receptor subtypes for agonist activity from 10 −3 to 10 −9 M. The results are tabulated as the mean EC 50 values from at least three independent experiments with the reported error being the standard error of the mean (SEM). A percentage denotes the percent maximal stimulatory response (compared to NDP-MSH) observed at 1 mM concentrations, but not enough stimulation was observed to determine an EC 50 value. The use of >1,000,000 indicates that the compound was examined but lacked agonist activity at concentrations up to 1 mM in at least two independent experiments. For partial agonist compounds (a sigmoidal dose-response curve was observed, but efficacy was less than 90% maximal NDP-MSH signal), compounds were binned into two groups: A (20-50% receptor activation) or B (51-90% receptor activation), Figure 2.

Structure-Activity Relationship Studies of 2718.003
The third compound examined, 2718.003, was previously reported to possess full mMC3R and mMC5R agonist activity (EC 50 = 350 and 540 nM, respectively), partially stimulated the mMC1R (50-90% at 100 µM concentrations) and the mMC4R (50% activation at 10 µM concentrations), and was an antagonist at the mMC4R (pA 2 = 5.8 with agonist MTII) [29]. In the present experiment (Table 4), full agonist activity was observed at the mMC3R (EC 50 = 2000 nM) and mMC5R (EC 50 = 1200 nM), as well as the mMC1R (EC 50 = 640 nM). The 6-fold decreased mMC3R potency between the previous reported values and the present study were not seen with the other two lead compounds, though the appearance of full MC1R agonism was also observed for the 2718.001 scaffold in the present experiment. This pyrrolidine bis-cyclic guanidine was a partial agonist at the mMC4R (EC 50 = 2500 nM) and possessed antagonist activity (pA 2 = 5.9 with agonist NDP-MSH) at the mMC4R. Similar to the other two scaffolds, this compound replicated as an mMC3R agonist and mMC4R partial activation/antagonist.
Truncation of the R1 R-isopropyl group to hydrogen (2718.029) resulted in an equipotent mMC1R, mMC3R, and mMC5R agonist compared to 2718.003, with similar partial agonist (EC 50 = 2500 nM) and antagonist (pA 2 = 5.8) activity at the mMC4R. The truncation of the R4 R-propyl group to hydrogen (2718.027) also maintained mMC1R, mMC3R, and mMC5R agonist activity and mMC4R antagonist activity relative to 2718.003. Partial agonism at the mMC1R and mMC5R was observed when the R3 R-cyclohexyl-methyl group was replaced with hydrogen (2718.028), which also partially activated the mMC3R and mMC4R. Similar partial agonism at the mMC1R was observed when the R5 position was shortened to ethyl (2718.014) or hydrogen (271.015), with minimal-to-no partial activation of the mMC3R, mMC4R, or mMC5R. Removal of both the R4 and R5 groups (replaced with hydrogen, 2718.016) produced an mMC1R partial agonist (EC 50 = 30,000 nM). This compound partially activated the mMC4R (25%) and mMC5R (55%), and was inactive at the mMC3R. Removal of the R1, R2, and R3 groups and amine backbone (resulting in monocyclic guanidine 2718.021) produced a full mMC5R agonist with 14-fold decreased potency relative to 2718.003. This compound partially activated the mMC1R, mMC3R, and mMC4R.     Inverting the R1 stereocenter in this scaffold to an S-isopropyl group (2718.053) resulted in the only stereochemical change that maintained mMC3R agonist potency relative to the corresponding lead compound. The 2718.053 ligand was an equipotent agonist (at the mMC1R, mMC3R, and mMC5R) and antagonist (mMC4R) compared to 2718.003. Changing the R2 (2718.052) and R4 (2718.050) stereochemistry resulted in equipotent mMC5R activity and decreased mMC1R (4-and 5-fold, respectively) potencies relative to 2718.003. Both changes resulted in partial agonists at the mMC3R and mMC4R. Inverting the R3 position (2718.051) decreased mMC1R and mMC5R agonist potencies 9-fold relative to 2718.003, resulted in partial agonist activity at the mMC4R, and partially activated the mMC3R. No double or triple stereochemical inversions were synthesized with this series. Inversion of the four stereochemical centers resulted in an equipotent mMC5R agonist (EC 50 = 3500 nM) relative to 2818.003, and partially activated the mMC1R, mMC3R, and mMC4R at 1 mM concentrations.

Discussion
The present SAR study was undertaken to identify the importance of sidechain functionality and stereochemistry at five positions within the reported mMC3R agonist pyrrolidine bis-cyclic guanidine scaffold [29]. By examining three different lead templates, the overall trends for the different positions could be identified.  Figure 5). Within this set of compounds, three (2718.024, 2718.026, and 2718.029) are derived from each lead ligand with the R1 functionalities replaced with hydrogen. Another three (2718.022, 2718.025, and 2718.027) are derived from each lead ligand with the corresponding R4 functionalities replaced with hydrogen. Analyzing three distinct molecular leads within this scaffold series identifies that the R1 and R4 positions may be removed while retaining full efficacy, suggesting that the double truncation compounds may be the minimal full agonist pharmacophores at the mMC3R, albeit with decreased potency compared to the starting ligands. When comparing the scaffolds (Figure 1), the R1 and R4 are constant between 2718.001 and 2718.002 (R-isobutyl and (S,S)-1-hydroxyethyl, respectively), and different compared to 2718.003 (R-isopropyl and R-propyl). If these groups are removed ( Figure 5 inset), the remaining core has common R2 (S-pyrrolidine) and R5 (4-tbutyl-cyclohexyl-methyl) groups, and cyclic R3 groups (R-benzyl for 2718.001 and R-cyclohexyl-methyl for 2718.002 and 2718.003). The similar substructure found for the full mMC3R agonists suggests this unit may be the minimal fragment required for full MC3R activity.
Compounds that possessed full agonist efficacy at the mMC3R were also full agonists at the mMC1R and mMC5R. Within the first template (2718.001), truncation of the R1 (2718.024) or R4 (2718.022) sidechains resulted in compounds possessing nanomolar EC 50 values at the mMC1R and mMC5R. Similar agonist potencies at the mMC1R and mMC5R were also observed for this template when the R1 (2718.033), R2 (2718.032), and R1/R2 (2718.040) positions were inverted. Changing the R2/R3 (2718.038) and R2/R4 (2718.036) stereocenters resulted in nanomolar mMC1R agonist compounds. A similar pattern was observed for truncation and stereochemical inversions in the second template (2718.002). Truncation of the R1 (2718.026) and R4 (2718.025) positions or inverting the R1 (2718.048) and R2 (2718.047) sidechains resulted in nanomolar agonist potencies at the mMC1R and mMC5R. From the third template compound, truncation of the R4 position (2718.027) was the only in this series that possessed nanomolar agonist potency at the mMC1R and mMC5R. In comparison, no compounds were observed to possess full agonist efficacy at the mMC4R. Overall, the pyrrolidine bicyclic guanidine templates examined herein appear to possess some agonist selectivity for the mMC3R over the mMC4R. Additional substitutions that decreased mMC3R efficacy and/or potency were found to maintain mMC1R and mMC5R activity, suggesting potential substitution sites that may be exploited to generate novel compounds selective for the mMC1R and/or mMC5R. mMC1R and mMC5R activity, suggesting potential substitution sites that may be exploited to generate novel compounds selective for the mMC1R and/or mMC5R. Figure 5. Structures of the full mMC3R agonists derived from this parallel SAR study. In addition to the three starting compounds, compounds truncated at the R1 (blue) or R4 (grey) positions retained full agonist efficacy. One additional R1 stereocenter inversion (2718.053) also retained full MC3R efficacy. A proposed common minimal MC3R agonist pharmaphore is proposed in the inset, where R2 could be R-benzyl or R-cyclohexyl-methyl.
Previously, a mixture-based positional scan was used to identify the three lead mMC3R agonist compounds that were the basis of the parallel SAR studies that were the focus of the current experiments [29]. Several mixtures not initially selected for deconvolution have individual compounds represented in the current experiments, permitting a retrospective analysis on the deconvolution process and the selection criteria used in the initial deconvolution. The initial deconvolution was based upon agonist potency at the mMC3R and relative selectivity for the mMC3R over the mMC4R [29]. Substitution of hydrogen at the first position (R1 Gly) at the mixture level resulted in minimal agonist activity and no selectivity over the mMC4R [29]. Two of the individual compounds with hydrogen at the R1 position (2718.024 and 2718.026) were 10-fold less potent at the mMC3R than the respective lead compounds, supporting the mixture-based screening results. In the mixture screen, the mixture defined with hydrogen at the R4 position (R3 in the previously published table) was the fifth most active mixture and possessed some selectivity for the mMC3R over the mMC4R [29]. The individual compounds with hydrogen at this position were equipotent (2718.022 and 2718.027) or possessed 4-fold decreased Figure 5. Structures of the full mMC3R agonists derived from this parallel SAR study. In addition to the three starting compounds, compounds truncated at the R1 (blue) or R4 (grey) positions retained full agonist efficacy. One additional R1 stereocenter inversion (2718.053) also retained full MC3R efficacy. A proposed common minimal MC3R agonist pharmaphore is proposed in the inset, where R2 could be R-benzyl or R-cyclohexyl-methyl.
Previously, a mixture-based positional scan was used to identify the three lead mMC3R agonist compounds that were the basis of the parallel SAR studies that were the focus of the current experiments [29]. Several mixtures not initially selected for deconvolution have individual compounds represented in the current experiments, permitting a retrospective analysis on the deconvolution process and the selection criteria used in the initial deconvolution. The initial deconvolution was based upon agonist potency at the mMC3R and relative selectivity for the mMC3R over the mMC4R [29]. Substitution of hydrogen at the first position (R1 Gly) at the mixture level resulted in minimal agonist activity and no selectivity over the mMC4R [29]. Two of the individual compounds with hydrogen at the R1 position (2718.024 and 2718.026) were 10-fold less potent at the mMC3R than the respective lead compounds, supporting the mixture-based screening results. In the mixture screen, the mixture defined with hydrogen at the R4 position (R3 in the previously published table) was the fifth most active mixture and possessed some selectivity for the mMC3R over the mMC4R [29]. The individual compounds with hydrogen at this position were equipotent (2718.022 and 2718.027) or possessed 4-fold decreased potency (2718.025) at the mMC3R compared to their respective leads. The minimal loss of potency could be relatively observed in the functional activity at the mixture level. Inverting the chirality of the incorporated amino acids at the R1 [DLeu (R-isobutyl) and DVal (R-isopropyl) to Leu (S-isobutyl) and Val (S-isopropyl)], R3 [DPhe (R-benzyl) to Phe (S-benzyl)], and R4 [DThr ((S,S)-1-hydoxyethyl) and DNva (R-propyl) to Thr ((R,R)-1-hydroxyethyl) and Nva (S-propyl)] decreased activities in the corresponding mixtures [29]. Therefore, the general decreased mMC3R activities may not be surprising in the current experiments from the stereochemical inversions. Although the purpose of the current set of experiments was to examine truncation and chirality changes to known mMC3R lead compounds, the mixture-based positional scan previously performed generally predicted the observed activity trends at the different substitution positions.
While these experiments were carried out to identify the common core scaffold requirements for MC3R agonism, two smaller fragments were observed to fully activate the mMC5R ( Figure 6). Compounds 2718.020 and 2718.021 were constructed by truncation of the R1, R2, and R3 positions, resulting in fragments possessing molecular weights under 300 Da that were micromolar potent mMC5R agonists ( Figure 6). The core of these two compounds is a 5-member heterocyclic guanidine group. The previously reported MC1R/MC5R antagonist JNJ-10229570, possessing IC 50 values in the 200-300 nM range [43], has a 5-member heterocyclic amino-thiadiazole core. The presence of the heterocyclic core in these compounds suggests that this motif may be important in developing MC5R agonist ligands and may offer guidance in substitutions that result in more potent compounds. potency (2718.025) at the mMC3R compared to their respective leads. The minimal loss of potency could be relatively observed in the functional activity at the mixture level. Inverting the chirality of the incorporated amino acids at the R1 [DLeu (R-isobutyl) and DVal (R-isopropyl) to Leu (S-isobutyl) and Val (S-isopropyl)], R3 [DPhe (R-benzyl) to Phe (Sbenzyl)], and R4 [DThr ((S,S)-1-hydoxyethyl) and DNva (R-propyl) to Thr ((R,R)-1-hydroxyethyl) and Nva (S-propyl)] decreased activities in the corresponding mixtures [29]. Therefore, the general decreased mMC3R activities may not be surprising in the current experiments from the stereochemical inversions. Although the purpose of the current set of experiments was to examine truncation and chirality changes to known mMC3R lead compounds, the mixture-based positional scan previously performed generally predicted the observed activity trends at the different substitution positions.
While these experiments were carried out to identify the common core scaffold requirements for MC3R agonism, two smaller fragments were observed to fully activate the mMC5R ( Figure 6). Compounds 2718.020 and 2718.021 were constructed by truncation of the R1, R2, and R3 positions, resulting in fragments possessing molecular weights under 300 Da that were micromolar potent mMC5R agonists ( Figure 6). The core of these two compounds is a 5-member heterocyclic guanidine group. The previously reported MC1R/MC5R antagonist JNJ-10229570, possessing IC50 values in the 200-300 nM range [43], has a 5-member heterocyclic amino-thiadiazole core. The presence of the heterocyclic core in these compounds suggests that this motif may be important in developing MC5R agonist ligands and may offer guidance in substitutions that result in more potent compounds. well. The cells were stimulated with the addition of 5 µL of stimulation buffer containing compound or forskolin (10 −4 M) and incubated in the dark at room temperature for 2 h. Following stimulation, streptavidin donor beads (0.5 µg) and biotinylated cAMP (0.62 µmol) were added to the wells in a green light environment with 10 µL of lysis buffer (5 mM HEPES, 0.3% Tween-20, 0.1% BSA, pH = 7.4) and the plates were incubated in the dark at room temperature for an additional 2 h. Plates were read on a Enspire (PerkinElmer, Waltham, MA, USA) Alpha plate reader using a pre-normalized assay protocol (set by the manufacturer).

Data Analysis
The pA 2 and EC 50 values represent the mean of at least three independent experiments performed in duplicate replicates. The compounds that were not active in two independent agonist experiments at the concentrations assayed (>1 mM for agonist assays) were not furthered examined. The pA 2 and EC 50 estimates and associated standard errors (SEM) were determined by fitting the data to a nonlinear least-squares analysis using the PRISM program (version 4.0, GraphPad Inc., San Diego, CA, USA). The compounds were assayed as formic acid salts.

Conclusions
The present experiments examined parallel structural modifications of three previously reported pyrrolidine bis-cyclic guanidine templates that were full MC3R agonists. Truncation of the R1 or R4 positions in all compounds resulted in full MC3R efficacy, implying that the R1 and R4 positions may be removed while retaining activity. Truncation of the R2, R3, and R5 positions resulted in loss of MC3R efficacy, indicating that the minimal MC3R pharmacophore requires such substitutions in this scaffold. Fragments with molecular weights under 300 Da were also identified as possessing full MC5R agonist efficacy. These studies identified minimal MC3R and MC5R pharmacophores based upon cyclic guanidine scaffolds, which may be useful in developing potent, selective ligands that clarify the roles of these receptors in vivo and as therapeutic leads to treat a variety of conditions caused by dysregulation of the melanocortin receptors. Funding: This work has been supported by NIH Grants R01DK124504.

Institutional Review Board Statement: Not applicable.
Data Availability Statement: The authors are happy work with individuals who would like to view raw data files. Please contact the corresponding author at: chaskell@umn.edu.