Combining mass spectrometry and machine learning to discover bioactive peptides

Peptides play important roles in regulating biological processes and form the basis of a multiplicity of therapeutic drugs. To date, only about 300 peptides in human have confirmed bioactivity, although tens of thousands have been reported in the literature. The majority of these are inactive degradation products of endogenous proteins and peptides, presenting a needle-in-a-haystack problem of identifying the most promising candidate peptides from large-scale peptidomics experiments to test for bioactivity. To address this challenge, we conducted a comprehensive analysis of the mammalian peptidome across seven tissues in four different mouse strains and used the data to train a machine learning model that predicts hundreds of peptide candidates based on patterns in the mass spectrometry data. We provide in silico validation examples and experimental confirmation of bioactivity for two peptides, demonstrating the utility of this resource for discovering lead peptides for further characterization and therapeutic development.

Recombinant adiponectin does not lower plasma glucose in animal models of type 2 diabetes

Aims/Hypothesis

Several studies have shown that adiponectin can lower blood glucose in diabetic mice. The aim of this study was to establish an effective adiponectin production process and to evaluate the anti-diabetic potential of the different adiponectin forms in diabetic mice and sand rats.

Methods

Human high molecular weight, mouse low molecular weight and mouse plus human globular adiponectin forms were expressed and purified from mammalian cells or yeast. The purified protein was administered at 10–30 mg/kg i.p. b.i.d. to diabetic db/db mice for 2 weeks. Furthermore, high molecular weight human and globular mouse adiponectin batches were administered at 5–15 mg/kg i.p. b.i.d. to diabetic sand rats for 12 days.

Results

Surprisingly, none of our batches had any effect on blood glucose, HbA1c, plasma lipids or body weight in diabetic db/db mice or sand rats. In vitro biological, biochemical and biophysical data suggest that the protein was correctly folded and biologically active.

Conclusions/Interpretation

Recombinant adiponectin is ineffective at lowering blood glucose in diabetic db/db mice or sand rats.

Adiponectin promotes macrophage polarization toward an anti-inflammatory phenotype

It is established that the adipocyte-derived cytokine adiponectin protects against cardiovascular and metabolic diseases, but the effect of this adipokine on macrophage polarization, an important mediator of disease progression, has never been assessed. We hypothesized that adiponectin modulates macrophage polarization from that resembling a classically activated M1 phenotype to that resembling alternatively-activated M2 cells. Peritoneal macrophages and the stromal vascular fraction (SVF) cells of adipose tissue isolated from adiponectin knock-out mice displayed increased M1 markers, including tumor necrosis factor-alpha, interleukin-6, and monocyte chemoattractant protein-1 and decreased M2 markers, including arginase-1, macrophage galactose N-acetyl-galactosamine specific lectin-1, and interleukin-10. The systemic delivery of adenovirus expressing adiponectin significantly augmented arginase-1 expression in peritoneal macrophages and SVF cells in both wild-type and adiponectin knock-out mice. In culture, the treatment of macrophages with recombinant adiponectin protein led to an increase in the levels of M2 markers and a reduction of reactive oxygen species and reactive oxygen species-related gene expression. Adiponectin also stimulated the expression of M2 markers and attenuated the expression of M1 markers in human monocyte-derived macrophages and SVF cells isolated from human adipose tissue. These data show that adiponectin functions as a regulator of macrophage polarization, and they indicate that conditions of high adiponectin expression may deter metabolic and cardiovascular disease progression by favoring an anti-inflammatory phenotype in macrophages.

Mitochondrial uncouplers with an extraordinary dynamic range

We have discovered that some weak uncouplers (typified by butylated hydroxytoluene) have a dynamic range of more than 10(6) in vitro: the concentration giving measurable uncoupling is less than one millionth of the concentration causing full uncoupling. They achieve this through a high-affinity interaction with the mitochondrial adenine nucleotide translocase that causes significant but limited uncoupling at extremely low uncoupler concentrations, together with more conventional uncoupling at much higher concentrations. Uncoupling at the translocase is not by a conventional weak acid/anion cycling mechanism since it is also caused by substituted triphenylphosphonium molecules, which are not anionic and cannot protonate. Covalent attachment of the uncoupler to a mitochondrially targeted hydrophobic cation sensitizes it to membrane potential, giving a small additional effect. The wide dynamic range of these uncouplers in isolated mitochondria and intact cells reveals a novel allosteric activation of proton transport through the adenine nucleotide translocase and provides a promising starting point for designing safer uncouplers for obesity therapy.

Changes in intracellular cAMP reported by a Redistribution assay using a cAMP-dependent protein kinase-green fluorescent protein chimera

We report on a novel method to monitor changes in intracellular cAMP concentration ([cAMP]i) within intact living cells using a chimeric fusion of the catalytic subunit of cAMP-dependent protein kinase to green fluorescent protein (PKAcat-GFP). In stably transfected unstimulated fibroblasts, fusion protein fluorescence is highly concentrated in aggregates throughout the cytoplasm and absent in the nucleus. Elevation of [cAMP]i disperses GFP fluorescence from the cytoplasmic aggregates within minutes. Spot-photobleach measurements show that the rate of exchange of GFP-labeled catalytic subunits at these aggregates increases in proportion to [cAMP]i. For any given stimulus, the response curve for dispersal of GFP fluorescence from aggregates agrees closely with the increase in total [cAMP]i as measured by standard in vitro methods (SPA). The redistribution of fluorescence is completely reversible: reduction of [cAMP]i results in return of fluorescence to the cytoplasmic aggregates. Consistent behaviour of PKAcat-GFP is seen in different cell backgrounds. We demonstrate that PKA Redistribution assays are suitable for measurement of changes in [cAMP]i brought about by both Gs- and Gi-protein-coupled receptor stimulation as well as by inhibition of cAMP phosphodiesterases.

Iodo-resiniferatoxin, a new potent vanilloid receptor antagonist

The highly potent vanilloid receptor (VR) agonist resiniferatoxin has been radiolabeled with 125I, and the pharmacology to the cloned rodent VR, VR1, and the endogenous VR in rat spinal cord membranes has been characterized. [125I]RTX binding to human embryonic kidney 293 cells expressing VR1 was reversible and with high affinity (Kd = 4.3 nM) in an apparent monophasic manner. In rat spinal cord membranes, [125I]RTX bound with a similar high affinity (Kd = 4.2 nM) to a limited number of binding sites (Bmax = 51 +/- 8 fmol/mg of protein). The pharmacology of recombinant rodent VR1 and the endogenous rat VR1 was indistinguishable when measuring displacement of [125I]RTX binding (i.e., the following rank order of affinity was observed: RTX > I-RTX > olvanil > capsaicin > capsazepine). Capsaicin and RTX induced large nondesensitizing currents in Xenopus laevis oocytes expressing VR1 (EC50 values were 1300 nM and 0.2 nM, respectively), whereas I-RTX induced no current per se at concentrations up to 10 microM. However, I-RTX completely blocked capsaicin-induced currents (IC50 = 3.9 nM). In vivo, I-RTX effectively blocked the pain responses elicited by capsaicin (ED50 = 16 ng/mouse, intrathecally). The present study showed that I-RTX is at least 40-fold more potent than the previously known VR antagonist, capsazepine. Thus, I-RTX as well as its radiolabeled form, should be highly useful for further exploring the physiological roles of VRs in the brain and periphery.

Adenosine is an agonist of the growth hormone secretagogue receptor

Growth hormone secretagogues (GHSs) are synthetic compounds that induce GH release in several species, including man. The aim of the current study was to identify hypothalamic GHS receptor (GHS-R) agonists. This led to the discovery of adenosine as a GHS-R agonist. We demonstrate that adenosine as well as the A1 adenosine receptor agonist N6-R-phenylisopropyladenosine (R-PIA) induce calcium responses, with EC50 values of 50 nM and 0.5 nM, respectively, in cells which express recombinant human GHS-R. However, neither compound induces a calcium response in nontransfected cells. Binding experiments show that adenosine and the GHS compound MK-0677 bind to membranes from GHS-R expressing cells with nearly identical Bmax values (2.6 +/- 0.1 x 10(-10) mol/mg protein for adenosine and 2.0 +/- 0.3 x 10(-10) mol/mg protein for MK-0677). However, no binding to membranes from nontransfected cells could be detected. Furthermore, we show that the IC50 values for inhibition of the adenosine, R-PIA, and GHS induced calcium responses by the GHS-R antagonist [D-Arg1, D-Phe5, D-Trp7,9, D-Leu11]-substance P are similar. These findings strongly suggest that adenosine and R-PIA are agonists of the GHS-R. Interestingly, neither adenosine nor R-PIA were able to induce GH release from rat pituitary cells in vitro. The implications of the latter finding is discussed.

Simultaneous visualization of the translocation of protein kinase Calpha-green fluorescent protein hybrids and intracellular calcium concentrations

The alpha isoform of protein kinase C (PKCalpha) is a ubiquitous protein kinase, which, upon activation, translocates rapidly from the cytoplasm to the plasma membrane. To follow this translocation, PKCalpha was tagged with a highly fluorescent derivative of green fluorescent protein and stably expressed in baby hamster kidney cells overexpressing the muscarinic type 1 receptor. Addition of the agonist carbamylcholine triggered the onset of translocation within 1 s. Half-maximal and maximal translocation occurred after about 3 and 15 s respectively. Plasma membrane association of the fusion protein was transient and the protein returned to the cytoplasm within about 45 s. A high-resolution study showed an almost homogeneous cytoplasmic distribution of tagged PKCalpha in unstimulated cells and virtually complete translocation to the plasma membrane in response to the phorbol ester, PMA. Simultaneous visualization of intracellular calcium concentration ([Ca2+]i) and PKCalpha translocation in single cells showed a good correlation between these parameters at intermediate and high concentrations of agonist. At low agonist concentration, a small increase in [Ca2+]i was observed, without detectable translocation of PKCalpha. In contrast, PMA induced translocation of PKCalpha without any increase in [Ca2+]i. Neither cytochalasin D nor colcemid influenced the distribution or calcium-dependent translocation of tagged PKCalpha, indicating that PKCalpha translocation may be independent of both actin filaments and microtubules. The time course of PKCalpha translocation is compatible with diffusion of the protein from its cytoplasmic localization to the plasma membrane.

A pronounced thymic B cell deficiency in the spontaneously diabetic BB rat

In an attempt to elucidate the origin of the T cell lymphopenia and/or the beta-cell-specific autoimmunity observed in diabetes-prone Bio-Breeding (DP-BB) rats, a thymic cDNA library was subjected to differential screening with thymic cDNA probes of DP-BB rats and nonlymphopenic nondiabetic controls. This approach resulted in the identification of a prominent lack of thymic B cells in DP-BB rats. This deficiency is distinct from a less pronounced peripheral B cell deficiency of different timing. The thymic B cell defect is linked to the lymphopenia trait on chromosome 4 and thereby with susceptibility to diabetes in crosses involving the DP-BB rat. In conclusion, our data suggest that the contribution of thymic B cells to the (negative) selection of thymocytes is inadequate in DP-BB rats, thus providing a plausible explanation for at least some of the spontaneous autoimmune phenomena in this animal model.

A pronounced thymic B cell deficiency in the spontaneously diabetic BB rat

In an attempt to elucidate the origin of the T cell lymphopenia and/or the beta-cell-specific autoimmunity observed in diabetes-prone Bio-Breeding (DP-BB) rats, a thymic cDNA library was subjected to differential screening with thymic cDNA probes of DP-BB rats and nonlymphopenic nondiabetic controls. This approach resulted in the identification of a prominent lack of thymic B cells in DP-BB rats. This deficiency is distinct from a less pronounced peripheral B cell deficiency of different timing. The thymic B cell defect is linked to the lymphopenia trait on chromosome 4 and thereby with susceptibility to diabetes in crosses involving the DP-BB rat. In conclusion, our data suggest that the contribution of thymic B cells to the (negative) selection of thymocytes is inadequate in DP-BB rats, thus providing a plausible explanation for at least some of the spontaneous autoimmune phenomena in this animal model.

Membrane anchoring of the autoantigen GAD65 to microvesicles in pancreatic beta-cells by palmitoylation in the NH2-terminal domain

Pancreatic beta-cells and gamma-aminobutyric acid (GABA)-secreting neurons both express the enzyme glutamic acid decarboxylase (GAD) which is a major target of autoantibodies associated with beta-cell destruction and impairment of GABA-ergic neurotransmitter pathways. The predominant form of GAD in pancreatic beta-cells, GAD65, is synthesized as a soluble hydrophilic molecule, which is modified to become firmly membrane anchored. Here we show by immunogold electron microscopy that GAD65 is localized to the membrane of small vesicles which are identical in size to small synaptic-like microvesicles in pancreatic beta-cells. The NH2-terminal domain of GAD65 is the site of a two-step modification, the last of which results in a firm membrane anchoring that involves posttranslational hydroxylamine sensitive palmitoylation. GAD65 can be released from the membrane by an apparent enzyme activity in islets, suggesting that the membrane anchoring step is reversible and potentially regulated. The hydrophobic modifications and consequent membrane anchoring of GAD65 to microvesicles that store its product GABA may be of functional importance and, moreover, significant for its selective role as an autoantigen.

A high-affinity uptake system for branched-chain amino acids in Saccharomyces cerevisiae

In order to isolate mutants with impaired uptake of branched-chain amino acids, mutants were induced which on complex medium were sensitive to an inhibitor of branched-chain amino acid biosynthesis. Eighteen mutants of independent origin were found. Ten of them were assayed for branched-chain amino acid uptake. Three of these were impaired in the uptake of L-valine, L-isoleucine and L-leucine, while the rest were unaffected in uptake of any of the three amino acids. Kinetics of the uptake by one selected mutant and the parental strain S288C were compared to models for one or two systems obeying Michaelis-Menten kinetics. This analysis suggested that a high-affinity system for all three amino acids is absent in the mutant, whereas low-affinity uptake of L-isoleucine and L-leucine by one or more systems remains unaffected. Moreover, medium-affinity uptake components for L-valine and L-leucine, not originally seen in the wild type, were identified in the mutant. In the wild type, 10 mM of any of the amino acids L-alanine, L-cysteine, L-isoleucine, L-leucine, L-tryptophan and L-valine reduce uptake of any of the three branched-chain amino acids. We propose that a permease responsible for high-affinity uptake of the branched-chain amino acids in strain S288C is partially or completely inactive in the mutant. Tetrad analysis shows that the phenotype can be ascribed to a single Mendelian gene. The wild-type allele is denoted BAP1 for branched-chain amino acid permease. The BAP1-dependent system is different from the general amino acid permease.