Analysis of longitudinal semicontinuous data using marginalized two-part model.

BACKGROUND: Connective tissue growth factor (CTGF), is a secreted matricellular factor that has been linked to increased risk of cardiovascular disease in diabetic subjects. Despite the biological role of CTGF in diabetes, it still remains unclear how CTGF expression is regulated. In this study, we aim to identify the clinical parameters that modulate plasma CTGF levels measured longitudinally in type 1 diabetic patients over a period of 10 years. A number of patients had negligible measured values of plasma CTGF that formed a point mass at zero, whereas others had high positive values of CTGF that were measured on a continuous scale. The observed combination of excessive zero and continuous positively distributed non-zero values in the CTGF outcome is referred to as semicontinuous data. METHODS: We propose a novel application of a marginalized two-part model (mTP) extended to accommodate longitudinal semicontinuous data in which the marginal mean is expressed in terms of the covariates and estimates of their effect on the mean responses are generated. The continuous component is assumed to follow distributions that stem from the generalized gamma family whereas the binary measure is analyzed using logistic model and both have correlated random effects. Other approaches including the one- and two-part with uncorrelated and correlated random effects models were also applied and their estimates were all compared. RESULTS: Our results using the mTP model identified intensive glucose control treatment and smoking as clinical factors that were associated with decreased and increased odds of observing non-zero CTGF values respectively. In addition, hemoglobin A1c, systolic blood pressure, and high density lipoprotein were all shown to be significant risk factors that contribute to increasing CTGF levels. These findings were consistently observed under the mTP model but varied with the distributions for the other models. Accuracy and precision of the mTP model was further validated using simulation studies. CONCLUSION: The mTP model identified new clinical determinants that modulate the levels of CTGF in diabetic subjects. Applicability of this approach can be extended to other biomarkers measured in patient populations that display a combination of negligible zero and non-zero values.

Plasma Connective Tissue Growth Factor (CTGF/CCN2) Levels Predict Myocardial Infarction in the Veterans Affairs Diabetes Trial (VADT) Cohort.

OBJECTIVE: Connective tissue growth factor (CTGF), also known as CCN2, is a potent chemotactic and extracellular matrix-inducing matricellular protein that has been implicated in progression of inflammatory and fibroproliferative disorders. An emerging role of CTGF/CCN2 is that of a prosclerotic factor implicated in the development of cardiac disease. Our objective was to determine the role of CTGF/CCN2 as a predictor of cardiovascular events in type 2 diabetes in the Veterans Affairs Diabetes Trial (VADT) cohort. RESEARCH DESIGN AND METHODS: Levels of CTGF/CCN2 were measured in 952 VADT patients a median of 1.9 years after entry into the study. Participants were followed for an average of 3.3 years for vascular outcomes. CTGF/CCN2 categories were defined as below the detectable limit (referent, 54.5%), lower half of detectable values (22.8%), and upper half of detectable values (22.7%). Hazard ratios (HRs) for cardiovascular end points in relation to CTGF/CCN2 categories were calculated by Cox proportional hazards models. RESULTS: During follow-up, 4.8% had a myocardial infarction (MI), 6.9% had an MI or cardiovascular death, and 6.9% died. After adjustments by conventional risk factors, individuals in the highest category of CTGF/CCN2 were at higher risk of MI (HR 2.43 [95% CI 1.15, 5.14]), MI or cardiovascular death (HR 2.71 [95% CI 1.44, 5.08]), and all-cause mortality (HR 2.70 [95% CI 1.43, 5.08]) relative to individuals with CTGF below the detectable limit. CONCLUSIONS: Our study indicates that high levels of CTGF/CCN2 predict future MI and cardiovascular death in patients with type 2 diabetes.

Multivariate Generalized Linear Mixed Models With Random Intercepts To Analyze Cardiovascular Risk Markers in Type-1 Diabetic Patients.

Statistical approaches tailored to analyzing longitudinal data that have multiple outcomes with different distributions are scarce. This paucity is due to the non-availability of multivariate distributions that jointly model outcomes with different distributions other than the multivariate normal. A plethora of research has been done on the specific combination of binary-Gaussian bivariate outcomes but a more general approach that allows other mixtures of distributions for multiple longitudinal outcomes has not been thoroughly demonstrated and examined. Here we study a multivariate generalized linear mixed models approach that jointly models multiple longitudinal outcomes with different combinations of distributions and incorporates the correlations between the various outcomes through separate yet correlated random intercepts. Every outcome is linked to the set of covariates through a proper link function that allows the incorporation and joint modelling of different distributions. A novel application was demonstrated on a cohort study of Type 1 diabetic patients to jointly model a mix of longitudinal cardiovascular outcomes and to explore for the first time the effect of glycemic control treatment, plasma prekallikrein biomarker, gender and age on cardiovascular risk factors collectively.

Relationship between vitamin D status and incidence of vascular events in the Veterans Affairs Diabetes Trial.

OBJECTIVE: Few studies have examined the relationship between vitamin D levels and incident cardiovascular events in large well-characterized patient cohorts. Therefore, our objective was to determine if low vitamin D levels predicted vascular complications of diabetes. METHODS: Prospective analysis of 936 veterans with type 2 diabetes (59.7 ± 8.4 years, 96.9% male) who participated in the Veteran Affairs Diabetes Trial (VADT) was conducted. 25(OH)-vitamin D was measured a median of two years after entry and participants were subsequently followed 3.7 years. Hazard ratios (HRs) were calculated for cardiovascular endpoints in relation to 25(OH)-vitamin D quartile. For microvascular endpoints, logistic regression was used to calculate odds ratios. RESULTS: After adjusting for age, minority status, treatment arm and history of prior event, individuals in the lowest vitamin D quartile (i.e., 1-15.9 ng/ml) were at similar risk of MI [HR = 1.13 (95% CI: 0.53, 2.42)], CHD [HR = 0.87 (95% CI: 0.49, 1.55)], congestive heart failure [HR = 1.44 (95% CI: 0.67, 3.06)], and death from any cause [HR = 1.04 (95% CI: 0.53, 2.04)] as individuals in the highest vitamin D quartile (i.e., 29.9-77.2 ng/ml). Similarly, there were no differences in the odds associated with retinopathy or renal disease onset or progression in the lowest versus highest vitamin D quartile. CONCLUSIONS: These data indicate that vitamin D status had no significant impact on the incidence of vascular events in a cohort of high-risk veterans with diabetes in which traditional risk factors were managed according to current treatment guidelines.

Biasing the parathyroid hormone receptor: relating in vitro ligand efficacy to in vivo biological activity.

Recent advances in our understanding of the pluridimensional nature of GPCR signaling have provided new insights into how orthosteric ligands regulate receptors, and how the phenomenon of functional selectivity or ligand "bias" might be exploited in pharmaceutical design. In contrast to the predictions of simple two-state models of GPCR function, where ligands affect all aspects of GPCR signaling proportionally, current models assume that receptors exist in multiple "active" conformations that differ in their ability to couple to different downstream effectors, and that structurally distinct ligands can bias signaling by preferentially stabilizing different active states. The type 1 parathyroid hormone receptor (PTH(1)R) offers unique insight into both the opportunities and challenges of exploiting ligand bias in pharmaceutical design, not only because numerous "biased" PTH analogs have been described but also because many of them have been characterized for biological activity in vivo. The PTH(1)R has pleiotropic signaling capacity, coupling to G(s), G(q/11), and G(i/o) family heterotrimeric G proteins, and binding arrestins, which mediate receptor desensitization and arrestin-dependent signaling. Here, we compare the activity of six different PTH(1)R ligands in a common HEK293 cell background using three readouts of receptor activation, cAMP production, intracellular calcium influx, and ERK1/2 activation, demonstrating the range of signal bias that can be experimentally observed in a "typical" screening program. When the in vitro activity profiles of these ligands are compared to their reported effects on bone mass in murine models, it is apparent that ligands activating cAMP production produce an anabolic response that does not correlate with the ability to also elicit calcium signaling. Paradoxically, one ligand that exhibits inverse agonism for cAMP production and arrestin-dependent ERK1/2 activation in vitro, (D-Trp(12), Tyr(34))-bPTH(7-34), reportedly produces an anabolic bone response in vivo despite an activity profile that is dramatically different from that of other active ligands. This underscores a major challenge facing efforts to rationally design "biased" GPCR ligands for therapeutic application. While it is clearly plausible to identify functionally selective ligands, the ability to predict how bias will affect drug response in vivo, is often lacking, especially in complex disorders.

The beta-arrestin pathway-selective type 1A angiotensin receptor (AT1A) agonist [Sar1,Ile4,Ile8]angiotensin II regulates a robust G protein-independent signaling network.

The angiotensin II peptide analog [Sar(1),Ile(4),Ile(8)]AngII (SII) is a biased AT(1A) receptor agonist that stimulates receptor phosphorylation, ß-arrestin recruitment, receptor internalization, and ß-arrestin-dependent ERK1/2 activation without activating heterotrimeric G-proteins. To determine the scope of G-protein-independent AT(1A) receptor signaling, we performed a gel-based phosphoproteomic analysis of AngII and SII-induced signaling in HEK cells stably expressing AT(1A) receptors. A total of 34 differentially phosphorylated proteins were detected, of which 16 were unique to SII and eight to AngII stimulation. MALDI-TOF/TOF mass fingerprinting was employed to identify 24 SII-sensitive phosphoprotein spots, of which three (two peptide inhibitors of protein phosphatase 2A (I1PP2A and I2PP2A) and prostaglandin E synthase 3 (PGES3)) were selected for validation and further study. We found that phosphorylation of I2PP2A was associated with rapid and transient inhibition of a ß-arrestin 2-associated pool of protein phosphatase 2A, leading to activation of Akt and increased phosphorylation of glycogen synthase kinase 3ß in an arrestin signalsome complex. SII-stimulated PGES3 phosphorylation coincided with an increase in ß-arrestin 1-associated PGES3 and an arrestin-dependent increase in cyclooxygenase 1-dependent prostaglandin E(2) synthesis. These findings suggest that AT(1A) receptors regulate a robust G protein-independent signaling network that affects protein phosphorylation and autocrine/paracrine prostaglandin production and that these pathways can be selectively modulated by biased ligands that antagonize G protein activation.

Plasma kallikrein promotes epidermal growth factor receptor transactivation and signaling in vascular smooth muscle through direct activation of protease-activated receptors.

The kallikrein-kinin system, along with the interlocking renin-angiotensin system, is a key regulator of vascular contractility and injury response. The principal effectors of the kallikrein-kinin system are plasma and tissue kallikreins, proteases that cleave high molecular weight kininogen to produce bradykinin. Most of the cellular actions of kallikrein (KK) are thought to be mediated by bradykinin, which acts via G protein-coupled B1 and B2 bradykinin receptors on VSMCs and endothelial cells. Here, we find that primary aortic vascular smooth muscle but not endothelial cells possess the ability to activate plasma prekallikrein. Surprisingly, exposing VSMCs to prekallikrein leads to activation of the ERK1/2 mitogen-activated protein kinase cascade via a mechanism that requires kallikrein activity but does not involve bradykinin receptors. In transfected HEK293 cells, we find that plasma kallikrein directly activates G protein-coupled protease-activated receptors (PARs) 1 and 2, which possess consensus kallikrein cleavage sites, but not PAR4. In vascular smooth muscles, KK stimulates ADAM (a disintegrin and metalloprotease) 17 activity via a PAR1/2 receptor-dependent mechanism, leading sequentially to release of the endogenous ADAM17 substrates, amphiregulin and tumor necrosis factor-α, metalloprotease-dependent transactivation of epidermal growth factor receptors, and metalloprotease and epidermal growth factor receptor-dependent ERK1/2 activation. These results suggest a novel mechanism of bradykinin-independent kallikrein action that may contribute to the regulation of vascular responses in pathophysiologic states, such as diabetes mellitus.

The adiponectin receptors AdipoR1 and AdipoR2 activate ERK1/2 through a Src/Ras-dependent pathway and stimulate cell growth.

Adiponectin is an adipocyte-derived cytokine that has attracted much attention because of its insulin-sensitizing effects in liver and skeletal muscle. Two adiponectin receptors, AdipoR1/R2, have been cloned, but relatively little is known about their intracellular signaling mechanisms. We found that full-length adiponectin rapidly and robustly activates the ERK1/2 mitogen-activated protein kinase pathway in primary vascular smooth muscle, vascular endothelial cells, and hepatocytes. In a HEK293 cell model, we found that downregulating AdipoR1/R2 simultaneously, but not individually, by RNA interference attenuated adiponectin-induced ERK1/2 activation, suggesting that either receptor was sufficient to mediate the response. Downregulation of T-cadherin, another adiponectin binding protein, enhanced the response. Downregulation of APPL1, an adapter protein and putative mediator of AdipoR1/R2 signaling, impaired adiponectin-stimulated ERK1/2 activation. Inhibiting PKA modestly attenuated ERK1/2 activation, while inhibition of Src family tyrosine kinases with PP2 abolished the response. The small GTPase inhibitor Clostridium difficile toxin B also produced complete inhibition. Adiponectin caused rapid, PP2-sensitive activation of Ras, but not the cAMP-regulated small GTPase, Rap1, suggesting that Src-dependent Ras activation is the dominant mechanism of adiponectin-stimulated ERK1/2 activation. To test whether Ras-ERK1/2 signaling by adiponectin was physiologically relevant, we determined the effects of overexpressing AdipoR1, adiponectin, or both on the rate of HEK293 cell growth. Overexpression of adiponectin alone, but not AdipoR1 alone, supported growth under serum-free conditions, while simultaneous expression of both led to further enhancement. These results suggest that adiponectin can exert proliferative effects by activating Ras signaling pathways.

Discovery of 5-[[4-[(2,3-dimethyl-2H-indazol-6-yl)methylamino]-2-pyrimidinyl]amino]-2-methyl-benzenesulfonamide (Pazopanib), a novel and potent vascular endothelial growth factor receptor inhibitor.

Inhibition of the vascular endothelial growth factor (VEGF) signaling pathway has emerged as one of the most promising new approaches for cancer therapy. We describe herein the key steps starting from an initial screening hit leading to the discovery of pazopanib, N(4)-(2,3-dimethyl-2H-indazol-6-yl)-N(4)-methyl-N(2)-(4-methyl-3-sulfonamidophenyl)-2,4-pyrimidinediamine, a potent pan-VEGF receptor (VEGFR) inhibitor under clinical development for renal-cell cancer and other solid tumors.

Role of beta-arrestin-mediated desensitization and signaling in the control of angiotensin AT1a receptor-stimulated transcription.

Heptahelical G protein-coupled receptors employ several mechanisms to activate the ERK1/2 cascade and control gene transcription. Previous work with the angiotensin AT1a receptor has shown that G(q/11) activation leads to a rapid and transient rise in ERK1/2 activity, whereas beta-arrestin binding supports sustained ERK1/2 activation by scaffolding a Raf.MEK.ERK complex associated with the internalized receptor. In this study, we compared the role of the two beta-arrestin isoforms in AT1a receptor desensitization, ERK1/2 activation and transcription using selective RNA interference. In HEK293 cells, both the native AT1a receptor and a G protein-coupling deficient DRY/AAY mutant recruited beta-arrestin1 and beta-arrestin2 upon angiotensin binding and internalized with the receptor. In contrast, only beta-arrestin2 supported protein kinase C-independent ERK1/2 activation by both the AT1a and DRY/AAY receptors. Using focused gene expression filter arrays to screen for endogenous transcriptional responses, we found that silencing beta-arrestin1 or beta-arrestin2 individually did not alter the response pattern but that silencing both caused a marked increase in the number of transcripts that were significantly up-regulated in response to AT1a receptor activation. The DRY/AAY receptor failed to elicit any detectable transcriptional response despite its ability to stimulate beta-arrestin2-dependent ERK1/2 activation. These results indicate that the transcriptional response to AT1a receptor activation primarily reflects heterotrimeric G protein activation. Although beta-arrestin1 and beta-arrestin2 are functionally specialized with respect to supporting G protein-independent ERK1/2 activation, their common effect is to dampen the transcriptional response by promoting receptor desensitization.

Pharmacokinetic-pharmacodynamic correlation from mouse to human with pazopanib, a multikinase angiogenesis inhibitor with potent antitumor and antiangiogenic activity.

With the development of targeted therapeutics, especially for small-molecule inhibitors, it is important to understand whether the observed in vivo efficacy correlates with the modulation of desired/intended target in vivo. We have developed a small-molecule inhibitor of all three vascular endothelial growth factor (VEGF) receptors (VEGFR), platelet-derived growth factor receptor, and c-Kit tyrosine kinases, pazopanib (GW786034), which selectively inhibits VEGF-induced endothelial cell proliferation. It has good oral exposure and inhibits angiogenesis and tumor growth in mice. Because bolus administration of the compound results in large differences in C(max) and C(trough), we investigated the effect of continuous infusion of a VEGFR inhibitor on tumor growth and angiogenesis. GW771806, which has similar enzyme and cellular profiles to GW786034, was used for these studies due to higher solubility requirements for infusion studies. Comparing the pharmacokinetics by two different routes of administration (bolus p.o. dosing and continuous infusion), we showed that the antitumor and antiangiogenic activity of VEGFR inhibitors is dependent on steady-state concentration of the compound above a threshold. The steady-state concentration required for these effects is consistent with the concentration required for the inhibition of VEGF-induced VEGFR2 phosphorylation in mouse lungs. Furthermore, the steady-state concentration of pazopanib determined from preclinical activity showed a strong correlation with the pharmacodynamic effects and antitumor activity in the phase I clinical trial.

Constitutive ERK1/2 activation by a chimeric neurokinin 1 receptor-beta-arrestin1 fusion protein. Probing the composition and function of the G protein-coupled receptor "signalsome".

The beta-arrestins, a small family of G protein-coupled receptor (GPCR)-binding proteins involved in receptor desensitization, have been shown to bind extracellular signal-regulated kinases 1 and 2 (ERK1/2) and function as scaffolds for GPCR-stimulated ERK1/2 activation. To better understand the mechanism of beta-arrestin-mediated ERK1/2 activation, we compared ERK1/2 activation by the wild-type neurokinin 1 (NK1) receptor with a chimeric NK1 receptor having beta-arrestin1 fused to the receptor C terminus (NK1-betaArr1). The NK1 receptor couples to both G(s) and G(q/11), resides on the plasma membrane, and mediates rapid ERK1/2 activation and nuclear translocation in response to neurokinin A. In contrast, NK1-betaArr1 is a G protein-uncoupled "constitutively desensitized" receptor that resides almost entirely in an intracellular endosomal compartment. Despite its inability to respond to neurokinin A, we found that NK1-betaArr1 expression caused robust constitutive activation of cytosolic ERK1/2 and that endogenous Raf, MEK1/2, and ERK1/2 coprecipitated in a complex with NK1-betaArr1. While agonist-dependent ERK1/2 activation by the NK1 receptor was independent of protein kinase A (PKA) or PKC activity, NK1-betaArr1-mediated ERK1/2 activation was completely inhibited when basal PKA and PKC activity were blocked. In addition, the rate of ERK1/2 dephosphorylation was slowed in NK1-betaArr1-expressing cells, suggesting that beta-arrestin-bound ERK1/2 is protected from mitogen-activated protein kinase phosphatase activity. These data suggest that beta-arrestin binding to GPCRs nucleates the formation of a stable "signalsome" that functions as a passive scaffold for the ERK1/2 cascade while confining ERK1/2 activity to an extranuclear compartment.

Discovery and evaluation of 2-anilino-5-aryloxazoles as a novel class of VEGFR2 kinase inhibitors.

A series of derivatives of 2-anilino-5-phenyloxazole (5) has been identified as inhibitors of VEGFR2 kinase. Herein we describe the structure-activity relationship (SAR) of this novel template. Optimization of both aryl rings led to very potent inhibitors at both the enzymatic and cellular levels. Oxazole 39 had excellent solubility and good oral PK when dosed as the bis-mesylate salt and demonstrated moderate in vivo efficacy against HT29 human colon tumor xenografts. X-ray crystallography confirmed the proposed binding mode, and comparison of oxazoles 39 and 46 revealed interesting differences in orientation of 2-pyridyl and 3-pyridyl rings, respectively, attached at the meta position of the 5-phenyl ring.

Not so strange bedfellows: G-protein-coupled receptors and Src family kinases.

Src family nonreceptor tyrosine kinases are an integral component of the signal transduction apparatus employed by growth factor receptor tyrosine kinases. As such, their role in cellular growth control and malignant transformation has been the subject of intensive investigation. In contrast, classical G-protein-coupled receptor (GPCR) signaling involves activation of second messenger-regulated serine/threonine kinases or ion channels, and is primarily involved in neurotransmission and the short-term regulation of intermediary metabolism. Over the past decade, this strictly dichotomous model of transmembrane signaling has been challenged by the discovery that GPCRs also exert control over cellular growth, proliferation, and differentiation, and do so by stimulating tyrosine phosphorylation cascades. Several mechanisms, from the direct association of Src family kinases with GPCRs or receptor-associated proteins, to the transactivation of receptor tyrosine kinases and focal adhesion complexes by G-protein-mediated signals, permit GPCRs to activate Src family kinases. Conversely, Src activity plays a central role in controlling GPCR trafficking and effects on cell proliferation and cytoskeletal rearrangement. It is now clear that GPCRs and Src family kinases do not belong to separate, exclusive clubs. Rather, these strange bedfellows are intimately involved in multilayered forms of crosstalk that influence a host of cellular processes.

Signaling in time and space: G protein-coupled receptors and mitogen-activated protein kinases.

Because of their central role in the cellular response to growth factors, assays of MAP kinase activity are commonly used in pharmaceutical screening efforts aimed at detecting chemical modifiers of growth regulatory pathways. As our understanding of the complexity of signal transduction networks expands, however, it is becoming apparent that previously unappreciated temporal and contextual factors have profound effects on MAP kinase function. This is exemplified by recent studies of the regulation of the ERK1/2 MAP kinase cascade by GPCRs. Depending on receptor and cell type, GPCR stimulation of ERK1/2 can reflect a heterogenous array of signaling events. Activation of second messenger-dependent protein kinases and cross talk between GPCRs and receptor or nonreceptor tyrosine kinases can all induce ERK1/2 activation. Furthermore, a growing body of data indicates that the mechanism of ERK1/2 activation is a major determinant of ERK1/2 function. Activation of a nuclear pool of ERK1/2 as a consequence of cross talk between GPCRs and growth factor receptor tyrosine kinases may provide a mitogenic stimulus. In contrast, activation of ERK1/2 in localized pools on the membrane or confined to endosomal vesicles through the utilization of focal adhesions or beta-arrestins as "scaffolds" may spatially constrain ERK1/2 activity and favor the phosphorylation of nonnuclear ERK substrates. Findings such as these suggest that screening strategies that use single readouts of MAP kinase activity or function are likely to miss important signaling events, and point to the need for a multidimensional approach to MAP kinase-based screening efforts.