Development and Validation of a Drug Activity Biomarker that Shows Target Inhibition in Cancer Patients Receiving Enzastaurin, a Novel Protein Kinase C-β Inhibitor

The type C family of protein kinases (PKC) is an important component of cellular signal transduction pathways (1). PKC has 12 known isoforms that are found in nearly all mammalian cell types. Hyperglycemia-induced PKC activation has been shown in retinal, renal, and vascular tissues, generating a hypothesis suggesting that activation of PKC, especially the β and δ isoforms, may mediate many diabetic complications (2). Furthermore, overexpression of PKC β can lead to the overproduction of vascular endothelial growth factor, a key mediator of angiogenesis, as well as inhibition of apoptotic cell death (3). Thus, PKC-β is recognized as a significant target for cancer chemotherapy (4). Conventional cancer chemotherapeutics are cytotoxic and often are administered to patients based on a recognized maximum tolerated dose (i.e., a dose that has been shown to cause significant, but manageable, toxicity in patients). New approaches in cancer chemotherapy revolve around generation of molecularly targeted agents. These agents are more specific for tumor cells and supporting tissue as they block activity of inappropriately expressed or overexpressed molecules in malignancy. As these new agents need not be dosed to maximum tolerated dose, determining the correct human dose necessitates novel, more specific methods. One approach uses assays that directly or indirectly measure inhibition of the intended molecular target in preclinical and early-stage clinical trials. Such assays may be considered to be “drug activity” biomarkers. Appropriate drug activity biomarkers can test whether the in vitro activity of the molecule is reproduced in animal models or humans. It is believed that one reason why many drugs fail in phase III (efficacy) trials is because of administration of an inappropriate dose. Drug activity biomarkers, such as the method we describe here, can aid the selection of appropriate dose for efficacy trials. Evaluation of such biomarker results can provide valuable information on the pharmacodynamic profile of a drug much sooner than long-term measures of disease progression. The ultimate objective of a biomarker strategy is to conduct smaller, “smarter” clinical trials that will speed-up the drug development process and potentially lead to the selection of doses with greater clinical efficacy.

Enzastaurin, formerly LY317615 HCl, a potent, novel macrocyclic bisindolylmaleimide, disrupts the intrinsic phosphotransferase activity of PKC-β. Administration of enzastaurin decreases vascular endothelial growth factor levels in tumor-bearing mice and suppresses growth of human glioblastoma and colon carcinoma xenografts in mice (5). Enzastaurin has been well tolerated in early human testing (6, 7) and has shown early promising phase II results in treatment of high-grade glioma (8). In this report, we describe the development and clinical validation of a biomarker assay to assess intracellular phosphoprotein signaling in human peripheral blood leukocytes. Our study showed decreased PKC activity in peripheral blood monocytes taken from cancer patients treated with enzastaurin.

Patients. Patients had advanced nonhematologic malignancies and participated in a phase I single-agent dose escalation trial at The University of Texas M.D. Anderson Cancer Center (Houston, TX) or at Johns Hopkins Medical Center (Baltimore, MD; ref. 6) or in a phase I study of enzastaurin in combination with capecitabine at University of California at Los Angeles Medical Center or at the University of Colorado Cancer Center (Denver, CO). Written informed consent was obtained from all patients and healthy donors according to institutional, state, and federal guidelines. For flow cytometric assays, patients were identified by numerals in sequence according to sample receipt (single agent study) or letters (combination study) and normal donors were assigned numbers in sequence.

Collection of peripheral blood mononuclear cells. Venous blood was collected from healthy normal donors or clinical trial patients into Becton Dickinson Vacutainer CPT tubes containing sodium heparin anticoagulant (Becton Dickinson, Franklin Lakes, NJ). Tubes were centrifuged for 20 minutes at 1,500 × g at room temperature. Processed CPT tubes were inspected for presence of mononuclear cell layer, then mononuclear cells were resuspended in the plasma layer by inverting the tube three to four times. The processed CPT tubes were stored or shipped overnight at 4°C. CPT tubes are important because they permit isolation of a relatively pure population of peripheral blood mononuclear cell (PBMC) preserved in autologous plasma sample. The PBMCs are remixed into the supernatant plasma for shipment but remain isolated from red cells and neutrophils by a gel barrier. Before beginning the assay, cells/plasma were mixed well, passed through a 70 μmol/L cell strainer, and warmed to room temperature.

Cell line and reagents. Cell culture medium and buffers were purchased from Invitrogen (Carlsbad, CA). For in vitro PKC inhibition studies, enzastaurin analogue (Eli Lilly & Co., Indianapolis, IN) was used. The human histiocytic lymphoma cell line U937 (9) was purchased from the American Type Culture Collection (Manassas, VA) and maintained in RPMI 1640 with 10% fetal bovine serum. The cell staining and wash buffer (assay buffer) was Dulbecco's PBS with addition of fetal bovine serum to 5% v/v. Methanol (anhydrous), DMSO, anisomycin, forskolin, 3-isobutyl-1-methylxanthine, and phorbol myristate acetate (PMA) were purchased from Sigma (St. Louis, MO). Stock solutions of agonists were prepared in DMSO and aliquots stored frozen at −70°C. Cytofix buffer and FITC-labeled surface antigen antibodies were purchased from BD Biosciences (San Diego, CA). Anti-phospho-(Ser)-PKC substrate antibody (rabbit polyclonal) was purchased from Cell Signaling Technology (Beverly, MA). Phycoerythrin-labeled goat anti-rabbit IgG was purchased from Biosource International (Camarillo, CA).

Stimulation and staining of phospho-PKC substrates in U937 cells. Aliquots of cells in log phase growth were washed with PBS and resuspended at 2 million/mL in RPMI with 10% fetal bovine serum. An aliquot of cells was pretreated with enzastaurin for 1 hour at 37°C. Cell stimulation and phosphoprotein staining was done using a modification of the method of Chow et al. (10). Briefly, 90 μL cells were dispensed into 12 × 75 mm polypropylene tubes followed by 10 μL of 10× working dilutions of agonist. Cells were stimulated for 20 minutes at 37°C, followed by addition of 100 μL of Cytofix buffer (diluted 1:1 in PBS) and incubation at 37°C for 10 minutes. Samples were then transferred to an ice bath, 1 mL ice-cold methanol added, and tubes were vortexed and held at 4°C for at least 30 minutes. Samples were then washed (2 mL wash buffer was added, tubes were centrifuged at 650 × g for 5 minutes, and supernatant was aspirated) and resuspended in 100 μL anti-phospho (p)-PKC substrate antibody (diluted 1/100 in assay buffer) or rabbit IgG (2 μg/mL) for control. Samples were incubated with primary antibody for 30 to 120 minutes, washed, and then incubated with phycoerythrin-labeled anti-rabbit IgG at 2 μg/mL for 15 to 30 minutes. Samples were washed and cells resuspended in 300 μL Cytofix buffer diluted 1:4 in standard PBS.

Stimulation and staining of p-PKC substrates in human leukocytes. Aliquots of cells in plasma were pretreated with enzastaurin (as appropriate) for 30 minutes at 37°C. PBMC stimulation and phosphoprotein staining was similar to that of U937 cells with some modifications. First, surface marker antibody (10 μL/test) was dispensed into tubes, followed by cells/plasma (sample volumes ranging from 200 to 400 μL were used). PMA was added using either 10× or 100× working solutions and cells were stimulated for 20 minutes at 37°C. This was followed by a rapid hypotonic lysis step to remove contaminating RBC (2 mL ice-cold distilled water was added to tubes, tubes were vortexed, and 250 μL of 10× PBS were added). Samples were centrifuged at 650 × g for 3 minutes, supernatant was aspirated, and cells were resuspended in 100 μL Cytofix buffer diluted 1:1 with standard PBS. Cells were fixed at 37°C for 10 minutes, permeabilized with methanol, and stained for intracellular p-PKC substrates as described above for U937 cells. Note that PMA treatment was found to cause monocyte adherence and clumping in some donors, leading to reduced monocyte numbers for analysis. This can be overcome, without affecting phosphoprotein results, by adding a small amount of EDTA solution (10 mmol/L) to tubes and vortexing well just before fixation. PKC activation in some patient samples was assayed by single-concentration (400 nmol/L PMA) stimulation of PBMC as described above. For others, a PMA titration curve was done. In addition, to provide a reference inhibitory control, an aliquot of patient PBMC in plasma was spiked in vitro with enzastaurin. Both aliquots were incubated at 37°C for 1 hour before proceeding with stimulation and measurement of PKC activation.

Flow cytometry analysis. Light scatter (forward and 90°), FITC, and phycoerythrin fluorescence signals were collected in listmode format on a Coulter XL or Beckman-Coulter FC500 flow cytometer (Beckman-Coulter, Miami, FL). For monocyte analysis, 3,000 CD14⁺ events were routinely analyzed (with a minimum of 500 cells). The collected listmode files were deconvoluted using Winlist software (Verity Software House, Topsham ME). Single-color controls were used for color compensation in the Winlist program. The mean phycoerythrin fluorescence intensity (MFI) was computed for each sample using a CD14-FITC versus 90° scatter gate.

Enzastaurin pharmacokinetic assay. Heparinized blood samples (5 mL) were collected from patients for pharmacokinetic assessment at steady state (on D7 and D15 of enzastaurin therapy). High-performance liquid chromatography with mass spectrometry was used to detect enzastaurin and its metabolites in plasma (Advion BioSciences, Inc., Ithaca, NY). The lower limit of quantification of this assay was 0.50 ng/mL. Pharmacokinetic variables, such as C_max and AUC_0-24, were calculated using noncompartmental methods from the plasma concentration-time profiles of enzastaurin and its metabolites with WinNonlin Pro 3.1 (Pharsight, Mountain View, CA).

Data analysis. Percentage fluorescence inhibition was calculated using the following formula:

where MFI_T refers to MFI of an enzastaurin-treated sample, MFI_B refers to unstimulated (basal) sample, and MFI_M refers to maximal stimulation MFI (PMA but no in vitro enzastaurin spike). When PMA titration curves were done on normal donor and patient PBMC, the area under the titration curve was calculated according to the linear trapezoidal rule. This value was designated as the integrated PMA response (IPR). Enzastaurin inhibition of IPR was calculated according to the following formula: [(IPR_max − IPR_{enzastaurin-treated}) / (IPR_max)] × 100, where IPR_max refers to the IPR of the PMA-only titration curve and IPR_{enzastaurin-treated} refers to the IPR of the PMA titration curve in the presence of enzastaurin. Day-to-day reproducibility of normal donor results was tested using the intercept term from a random-effects model of the logarithm of the ratio of the two IPR values using JMP software (SAS Institute, Inc., Cary, NC) with REML option. Statistical significance of PKC inhibition by enzastaurin in the capecitabine combination study was determined by log transformation of data and analysis by a linear mixed effect model (SAS Institute).

Flow cytometric detection of PKC signaling in U937 cells. We characterized our p-PKC substrate antibody using U937 cells, a cell line that has been shown by other techniques to express the PKC βI and II isoforms, among others (11–13). Fluorescence histograms from a representative experiment (Fig. 1) show that basal p-PKC substrate signal measured was ∼15-fold greater than nonspecific binding to rabbit IgG (MFI = 62.0 for p-PKC substrate antibody versus 4.3 for rabbit IgG control). A 20-minute treatment with PMA at 500 nmol/L resulted in a 3-fold increase in p-PKC substrate signal; however, treatment with the mitogen-activated protein kinase agonist anigsomycin (1 μg/mL) did not increase signal. Likewise, treatment conditions used to stimulate PKA activity (20-minute incubation with 10 μmol/L forskolin and 500 μmol/L 3-isobutyl-1-methylxanthine) did not affect the p-PKC substrate signal. A 1-hour pretreatment with 2 μmol/L enzastaurin blocked PMA-induced PKC activation by 73%.

Comparison of PKC activation in peripheral blood leukocyte subsets shows that monocytes are best responders. Antibodies to distinct leukocyte surface antigens were added to PBMC samples at the time of PMA stimulation. The surface antigen antibodies chosen were FITC labeled (emission 525 nm). After sample stimulation (400 nmol/L PMA for 20 minutes) and permeabilization, binding of the p-PKC substrate antibody was detected by incubating the cells with (phycoerythrin)-labeled (emission 575 nm) anti-rabbit IgG antibody. This method provides for multicolor flow cytometric analysis of PKC activation of distinct leukocyte subsets in a mixed PBMC population. The representative data shown in Fig. 2 consists of flow cytometric dual-variable plots of membrane surface marker expression (X axis) versus p-PKC substrate reactivity (Y axis) for human blood that identifies monocytes (A-C) and human B lymphocytes (D-F). Although both cell populations showed robust PMA-induced phosphorylation of PKC substrates (PKC activation), the monocyte response was larger (10-fold window for monocytes versus 4-fold window for B cells). Both populations exhibited low basal levels of p-PKC substrates (Fig. 1A and D), but the monocytes had a larger and more uniform expression of p-PKC substrates than B cells after PMA stimulation (Fig. 1B and E). Importantly, for our purposes, enzastaurin achieved more complete inhibition of PKC activation in monocytes (Fig. 1C) compared with B lymphocytes (Fig. 1F). Table 1 summarizes the PMA-induced activation and enzastaurin inhibition for human CD4⁺ T lymphocytes, CD8⁺ T lymphocytes, CD14⁺ monocytes, and CD19⁺ B lymphocytes, and for all leukocytes (no specific surface antigen measured). Monocytes were chosen for further studies.

Single-agent clinical study: ex vivo assay reveals decreased PMA-induced PKC activity in patient monocytes following enzastaurin dose. Based on these in vitro results, blood samples were collected from patients enrolled in a phase I clinical trial. Blood was collected predose (within 4 hours of initial dose) and at steady-state drug levels that were achieved after 14 and 28 days of receiving once daily oral doses of 525 mg enzastaurin. PBMC specimens displayed good viability (trypan blue exclusion) after shipment (>90%, data not shown). Figure 3 shows CD14-gated p-PKC substrate MFI for samples from the four patients that were assayed ± PMA stimulation. For each blood collection day shown on the X axis, basal, PMA-treated, and enzastaurin-spiked/PMA-treated PKC activation results (tested in duplicate) are represented by distinct columns. Results from patients 2 through 4 indicate inhibition of PMA-induced PKC activity after initiation of the enzastaurin dose. Results from patient 1 indicate a lack of in vivo effect although the in vitro spike implied technical success of the assay.

Enzastaurin shifts PMA-induced PKC activity concentration response curves. We investigated the sensitivity of our monocyte PKC biomarker by adding varying dilutions of enzastaurin to 8 mL whole blood in six CPT Vacutainer tubes. Tubes were incubated for 1 hour at room temperature before centrifuging the tubes to obtain isolated PBMC. An in vitro PMA concentration response test was done on the resultant treated PBMC preparations. The MFI of p-PKC substrate signal versus PMA concentration was plotted for each in vitro enzastaurin dose and the area under each titration curve was calculated to be the integrated PMA response (IPR; described in Materials and Methods). The plot of enzastaurin concentration added to the whole blood sample versus IPR inhibition is shown in Fig. 4. PMA-induced PKC activation was inhibited in a concentration-dependent manner with an IC₅₀ calculated to be 1.2 μmol/L in human whole blood. Blood samples from four patients in the single-agent study were assayed for PKC activity using the IPR method. Samples were studied predose and after 14 days of once daily enzastaurin oral doses of 525 mg. Figure 5 shows MFI values for PKC activation after stimulation with PMA concentrations from 25 to 1,600 nmol/L. The small (∼3 mL) volume of plasma/cells obtained from each patient specimen allows for only single tube assay at each PMA concentration. Comparison of predose and day 14 PMA titration curves for these patients reveals distinct inhibition of PKC activation in patients dosed with enzastaurin.

Normal donor reproducibility. Blood samples were collected from 16 normal donors 7 days apart and tested in duplicate using the PMA titration-IPR method. A ratio of IPR values (draw 2/draw 1) was computed for each donor and summarized in Table 2. Excellent day-to-day reproducibility for the normal donor population was shown. The drift was tested using the intercept term from a random-effects model on the logarithm of the ratio of IPRs from draw 1 to draw 2 and was found to be nonsignificant (P = 0.075).

Phase I trial of enzastaurin in combination with capecitabine: ex vivo biomarker results indicate PKC inhibition in patients following enzastaurin dose. Additional investigation was done in a subsequent phase 1 study where patients received enzastaurin in combination with capecitabine. Blood samples for monocyte PKC activity assay were collected at predose, after 4 hours, and 14 days of single, daily 525 mg oral doses of enzastaurin (one patient received daily doses of 350 mg). The PMA-induced PKC activity was measured on the patient sample along with an approximate collection time-matched normal donor control sample (for assurance of assay quality). In this study, all appropriate PBMC samples were received in a timely manner for 9 of 17 study enrollees. Reasons for missing samples include not drawn (especially for baseline), sample handling (cells frozen, etc.), patient discontinued from trial, and late sample arrival (shipping problem). Previous normal donor in vitro studies indicated that basal PKC activity is increased in samples held for 48 hours compared with those held for 24 hours (data not shown). For clinical specimens, sample age was tracked and the results reported here represent specimens held from 21 to 29 hours. Assay runs were considered valid when the normal donor raw data indicated a minimal 8.26-fold increase in p-PKC substrate signal after treatment with 3,200 nmol/L PMA, a criterion derived from the 16-donor 2-week reproducibility study (Table 2). Patient PBMCs were stimulated with concentrations of PMA ranging from 100 to 3,200 nmol/L. The p-PKC substrate fluorescence results for the nine included patient samples were averaged and displayed in Fig. 6. The 4-hour and 14-day curves were statistically compared with the predose baseline using a linear mixed-effect model (SAS MIXED) with fixed effects for time and concentration and subject as a random effect. Using this analysis, both the 4-hour and 14-day values are significantly lower compared with the baseline values (P < 0.0001). Pharmacokinetic assessment in patients receiving the 525 or 500 mg enzastaurin dose indicated that the median time to maximum plasma concentration (C_max) at steady state for enzastaurin was 4.17 hours (ranging from 1.00 to 6.08 hours). The average plasma concentration and %CV at steady state for enzastaurin was 1,125.3 nmol/L (43.5%). One patient received 350 mg doses of enzastaurin and the average steady-state concentration for that patient was ∼300 nmol/L.

Protein kinases regulate almost all cellular processes and may exhibit aberrant expression or activity in various disease states. PKC is increasingly recognized as an important molecular target for intervention in diabetic microvascular complications and tumor angiogenesis. We have developed and applied a drug activity biomarker to detect PKC activation and inhibition in a surrogate cell, human monocytes. Our method is based on flow cytometric measurement of intracellular phosphorylated PKC substrates. Our data here shows that PKC activation was inhibited in cells after exposure (either in vitro or in vivo) to the PKC inhibitor enzastaurin.

Flow cytometric measurement of intracellular signaling using phosphospecific antibodies was first reported by Chow et al. (10). The authors showed strong activation of extracellular signal-regulated kinase in CD3⁺ lymphocytes following PMA or anti-CD3 stimulation and dose-dependent inhibition of phospho–extracellular signal-regulated kinase following in vitro treatment with an investigational Raf kinase inhibitor. Analysis of phosphoproteins by flow cytometry provides several advantages over Western blots, the standard technique used to detect phosphoepitopes. Chief among these is single cell analysis. Using flow cytometry, responses in minor cell subpopulations can be measured easily. Western blotting provides a single average value obtained from millions of lysed cells, thus hiding robust responses from small subpopulations. In addition, multivariable acquisition makes it possible to correlate multiple intracellular and extracellular markers simultaneously. Jacoberger et al., for example, examined cell cycle related distribution of phospho-STAT-5 in chronic myelogenous leukemia cell lines using probes for cellular DNA content and antibody to the mitotic marker MPM2 (14). Success of the flow cytometry technique is highly dependent on availability of quality reagent-grade antibodies that bind specifically to the phosphorylated (active) epitopes and do so in fixed, intact cells. Although this technique is in its infancy, vendors are now responding by developing and testing phosphospecific antibodies designed specifically for flow cytometry. Krutzik et al. (15) recently published an excellent review of the techniques and challenges involved in measuring protein phosphorylation at the single cell level. Detection of phosphoproteins by flow cytometry has been pushed to the current analytic limits by Perez and Nolan (16). Using a flow cytometer modified to collect signals from up to 11 different fluorochromes, the authors evaluated activity and kinetics of multiple kinase families in naïve and memory T lymphocytes.

In our studies, we first tested phospho–extracellular signal-regulated kinase and phospho–mitogen-activated protein/extracellular signal-regulated kinase kinase antibodies (data not shown). A number of commercially available anti-phospho-extracellular signal-regulated kinase antibodies can be used for robust signal by flow cytometry; however, we found that the best sensitivity to enzastaurin action was revealed by using an antibody to phosphorylated PKC substrates. The antibody recognizes phosphorylated serine residues in specific motifs (serine residues surrounded by lysine or arginine in positions −2 and +2 and a hydrophobic residue in position +1) of substrates of the classic PKC isoforms α, βI, βII, and γ (17). The antibody has been used for PKC studies with Western blot and ELISA readout (18–20). Although not specific for PKC βII, we were able to use this antibody to show enzastaurin inhibition of the PKC signaling cascade. Our studies revealed that monocytes, a minor population of human PBMC, are potent responders to PMA-induced PKC activation. Both monocytes and B-lymphocytes have been shown to have abundant PKC activity, particularly the β isoforms (21, 22). Although all cells responded to PMA to varying degrees, monocytes generated a distinctly better signal window. Flow cytometry permitted us to focus on the appropriate subset and improved the assay signal window and thus assay sensitivity.

Intracellular phosphoprotein assays at the single cell level should continue to make important contributions to our understanding of the complex signaling circuitry of cells. The early reports of this novel technique have focused on signaling in cell lines or PBMC treated in vitro. Our study is one of few to use ex vivo stimulation to glean pharmacodynamic information from patient samples. Our initial results suggest that the administered enzastaurin dose is active against its intended molecular target in humans, albeit in surrogate cells. The monocyte serves as a conveniently obtained specimen that is robustly stimulated and dramatically inhibited. Desplat et al. recently reported flow cytometry analysis of leukemic blasts for expression of intracellular phosphotyrosine. The assay was used to monitor patients on imitinab therapy. Interestingly, bone marrow cells from chronic myelogenous leukemia patients who progressed to blast crisis despite continuing imitinab therapy exhibited high constitutive phosphotyrosine signal that could not be reduced by ex vivo culture with imitinab (23). These results linked hyperphosphorylation in leukemic blasts to failed inhibition of the molecular target and to clinical relapse.

Our drug activity biomarker has provided one of the first glimpses of intended target inhibition by enzastaurin in humans. Two factors that contributed to our success were selection of an appropriate antibody and using monocytes as surrogate target cells. At the outset of our studies, we did not know how monocytes from cancer patients would perform in our PKC activity assay compared with normal donor monocytes. We increased our chances of detecting PMA and drug effects by testing a 32-fold concentration range of PMA, with this result being conveniently summarized using an area under curve calculation (IPR). Clinical use of molecularly targeted therapies, such as enzastaurin, may possibly be maximized by codevelopment of specific assays, similar to those described in this report, to assist clinicians in identifying appropriate dosage and monitoring efficacy, toxicity, and resistance to these novel agents. Despite intersubject variability in the magnitude of monocyte responses to PMA, our intrasubject study using normal donors showed excellent week-to-week reproducibility. This is an important feature because drug activity biomarkers are most informative when done on sequentially obtained patient samples taken before and during the course of drug administration. Our assay of monocyte PKC activity has provided valuable information for further development of enzastaurin, yet use of functional assays in the clinical trial setting presents certain unique challenges. Many clinical laboratory markers are measured on patient serum or plasma samples that may be frozen for later analysis. The logistics of transporting and maintaining viable leukocytes, together with the throughput of current methods, means that intracellular phosphoprotein assays are most suitable for small phase I trials.

Despite significant treatment advances in the last decade, patients with advanced or metastatic cancers still have poor prognoses. Enzastaurin is an exciting example of a molecularly targeted anticancer agent. Presumably, with such agents, physicians could use methods that evaluate specific target inhibition to determine effective dose. Enzastaurin blocks activity of PKC, a key enzyme in the vascular endothelial growth factor signaling cascade that can ultimately induce tumor angiogenesis. We have used a multiparameter, flow cytometry assay of peripheral blood monocytes to show PKC inhibition in these cells after patients received enzastaurin. Data obtained from these studies should be useful for selection of efficacy doses and design of future clinical evaluations.

We thank Dr. Michael Lahn for critical review of the manuscript, Christelle Darstein for statistical analysis, Lilly associates Kim Gill and Jennifer Hatmacher for managing clinical specimen collection and shipment, Dr. David Hedley for fostering our initial interest in intracellular phosphoprotein assays, and Dr. Garry Nolan for his encouraging endorsement of this work.