Human platelets express both cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2). Variation in COX-2 expression could be a mechanism for variable response to aspirin.
Human platelets express both cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2). Variation in COX-2 expression could be a mechanism for variable response to aspirin.
The hypotheses were that circulating canine platelets express COX-1 and COX-2, and that aspirin alters COX expression. The objective was to identify changes in platelet COX expression and in platelet function caused by aspirin administration to dogs.
Eight female, intact hounds.
A single population, repeated measures design was used to evaluate platelet COX-1 and COX-2 expression by flow cytometry before and after aspirin (10 mg/kg Q12h for 10 days). Platelet function was analyzed via PFA-100® (collagen/epinephrine), and urine 11-dehydro-thromboxane B2 (11-dTXB2) was measured and normalized to urinary creatinine. Differences in COX expression, PFA-100® closure times, and urine 11-dTXB2 : creatinine ratio were analyzed before and after aspirin administration.
Both COX-1 and COX-2 were expressed in canine platelets. COX-1 mean fluorescent intensity (MFI) increased in all dogs, by 250% (range 63–476%), while COX-2 expression did not change significantly (P = 0.124) after aspirin exposure, with large interindividual variation. PFA-100® closure times were prolonged and urine 11-dTXB2 concentration decreased in all dogs after aspirin administration.
Canine platelets express both COX isoforms. After aspirin exposure, COX-1 expression increased despite impairment of platelet function, while COX-2 expression varied markedly among dogs. Variability in platelet COX-2 expression should be explored as a potential mechanism for, or marker of, variable aspirin responsiveness.
fluorescence-activated cell sorting–phosphate buffered saline buffer
immune mediated hemolytic anemia
mean fluorescence intensity
platelet function analyzer
The cyclooxygenase (COX) enzyme plays a key role in normal platelet function. Platelet COX activity is required to convert arachidonic acid to prostaglandin H2, leading to the production of several biologically active prostaglandins, including thromboxane A2.[1, 2] Thromboxane A2 is a potent vasoconstrictor and platelet activator, and production of thromboxane A2 by platelets is essential for normal platelet function.
At least 3 COX isoforms have been described in dogs. COX-1 is constitutively expressed in most body systems, and functions to maintain normal cellular activities including platelet thromboxane A2 production, maintenance of gastric mucosal protection, and preservation of renal blood flow.[2, 3] The COX-3 isoform is transcribed from the COX-1 gene as a splice variant and has similar catalytic features, but only about 20% of the activity of COX-1.[4, 5] Relative to COX-1, COX-2 has a far more limited tissue distribution, and expression is often at much lower levels. COX-2 expression occurs in the brain, kidney, thymus, and the vascular endothelium.[2, 6-8] COX-2 is also present in circulating monocytes, tissue macrophages, and fibroblasts.[2, 8] COX-2 expression can be induced by various endogenous and exogenous mediators including inflammatory cytokines, growth factors, and endotoxin.[2, 3, 9] Induction of COX-2 occurs primarily at the post-transcriptional level.
Because mature platelets do not contain a nucleus they are therefore incapable of messenger ribonucleic acid (mRNA) synthesis, and because COX-1 is usually constitutively expressed by cells whereas COX-2 expression is more typically induced, it was long believed that COX-1 was the only COX isoform expressed in circulating platelets, and that COX induction could happen only at the level of the bone marrow. As a result, it was also believed that COX-1 was the primary mediator of platelet thromboxane A2 production.[7, 11-13] Recently, however, a flow cytometry technique has been described that has identified and quantified the expression of both COX-1 and COX-2 in some mature circulating human platelets. The COX-2 isoform has also been identified more frequently in young (reticulated) platelets.[1, 6, 14] These recent discoveries suggest that COX-2 could in fact play a role in platelet function.[8, 11] To the authors' knowledge, COX-2 has not previously been identified in circulating canine platelets.
Aspirin (acetylsalicylic acid) is one of the most commonly administered antiplatelet medications in both human and veterinary medicine. Aspirin binds to and irreversibly acetylates serine residues in the active sites of the COX-1 and COX-2 enzymes, resulting in decreased platelet thromboxane A2 production and inhibition of platelet function.[1, 15-20]
The main objective of this study was to utilize flow cytometry to identify and to quantify relative COX-1 and COX-2 expression in circulating canine platelets. To further investigate the respective roles of the COX isoforms, this study also evaluated the effects of aspirin on platelet expression of both isoforms. Platelet function analysis and measurement of urinary 11-dehydro-thromboxane B2 (11-dTXB2) were also employed to explore the relationships between COX isoform expression, thromboxane production, and platelet function in aspirin-treated dogs.
Eight healthy female intact Walker Hound dogs were used in the study. The dogs were not exposed to any drugs or vaccines for at least 2 months before study commencement. Aspirin1 was administered to each dog at a standard anti-inflammatory dose of 10 mg/kg PO every 12 hours for 10 days. The administered aspirin dose was 10.62 ± 0.610 (mean, SD) mg/kg, PO, every 12 hours. All dogs completed the study. Blood was collected for flow cytometric measurement of platelet COX expression before the first aspirin dose, 1–4 hours after aspirin administration on Day 10, and 14 days after the last dose of aspirin (washout). Blood and urine were collected for platelet function analyzer PFA-100® testing and urinary 11-dTXB2 analysis at 2 time points, before the 1st aspirin dose and 1–4 hours after aspirin administration on Day 10.
The mean age of the dogs was 5 years (range, 1–6 years). Individual dog weights ranged from 20.5 to 24.2 kg (median, 22.3 kg). Body weight was obtained at the beginning of the study and used to calculate all subsequent dosing. Normal health status was established based on the normal results of physical examination, buccal mucosal bleeding time, complete blood count (including manual platelet count), serum biochemistry, urinalysis, prothrombin time, partial thromboplastin time, von Willebrand factor antigen testing (ELISA method), heartworm testing, and rickettsial and Babesia serology. Animal use was approved by the Mississippi State University Institutional Animal Care and Use Committee.
Blood was collected via jugular venipuncture with a 20 gauge needle into a glass vacutainer tube containing 3.8% sodium citrate.2 Sample preparation for flow cytometry was initiated within 1 hour of collection.
A modification of a previously described protocol for quantitating COX-1 expression on human platelets was used to label COX-1. Briefly, 5 μL of citrated whole blood was added to 45 μL of fluorescence-activated cell sorting—phosphate buffered saline buffer (FACS-PBS) and incubated for 30 minutes with 10 μL of fluorescein isothiocyanate (FITC)-conjugated monoclonal mouse anti-ovine-COX-1 antibody.3 Five microliters of IgG2b-FITC isotype control4 was run in parallel with each set of samples. The samples were then incubated for 30 minutes with 3 μL of monoclonal antibody mouse anti-human CD9 : R-phycoerythrin (RPE) antibody5 to label platelets. Samples were fixed for 10 minutes at 4°C in the dark with 1% paraformaldehyde.6
A modification of a previously described protocol for quantitating COX-2 expression in permeabilized human platelets was used to label COX-2. Briefly, 5 μL of citrated whole blood was added to 45 μL of FACS-PBS and fixed at 4°C in the dark for 10 minutes with 1% paraformaldehyde. Samples were washed with phosphate buffered saline (PBS) and pelleted by centrifugation (400 × g for 7 minutes). The supernatant was discarded, and the pellet was resuspended and incubated in 0.3% Triton X-1007 for 10 minutes, and the wash step was repeated. The samples were then incubated at room temperature in the dark for 30 minutes with 15 μL of a monoclonal FITC-conjugated mouse anti-human-COX-2 antibody.8 Platelets were labeled by adding 20 μL of monoclonal mouse anti-pig CD61-purified antibody9 and incubated for 30 minutes, followed by the addition of 100 μL of goat anti-mouse immunoglobulin G (IgG):RPE antibody10 and incubated for 30 minutes. A FITC-IgG1 isotype control11 was run in parallel with all samples.
After labeling of platelets, samples were stored in the dark at 4°C until analysis. Flow cytometric analysis was performed within 2 hours of sample preparation. Platelets were analyzed using a flow cytometer12 at a wavelength of 488 nanometers with CellQuest Pro software.13 Platelet populations were displayed on log forward-scatter versus log side-scatter plots. Compensation was performed to compensate for spectral overlap of 2 different fluorochromes, FITC and RPE. Gates were adjusted to baseline platelet populations. A total of 5,000 gated events were recorded for both COX isoforms. COX expression was quantified by the intensity of antibody staining and was expressed as MFI. Fluorescence from the isotype control was subtracted from each sample. A histogram was created with MFI on the x-axis and number of events on the y-axis.
All the antibodies used in this study were cross reactive with canine antigens, according to the manufacturers. Previous studies have demonstrated that all the critical catalytic residues found in ovine COX-1 are also found in the canine COX-1 molecule, and that there is a greater than 90% sequence similarity between the human and canine COX-2 molecule. Furthermore, anti-ovine and anti-human antibodies have been used previously for identification of canine COX-1 and COX-2, respectively. In pilot studies in this laboratory before initiation of this study, the expression of both COX isoforms was analyzed in canine platelets via flow cytometry with both platelet-rich plasma and whole blood samples, and samples were also analyzed with several antibodies (anti-CD9, anti-CD61, and anti-CD62P) known to be expressed on the surface of platelets. The sample type and antibody which provided the most consistent identification of the platelet population for each isoform assay was then used in the study.
Platelet function was analyzed using a commercial point-of-care PFA-100®.14 The PFA-100® has been previously evaluated for use in dogs, and the instrument was used according to manufacturer's instructions.[22-24] Briefly, the PFA-100® is an in vitro instrument that measures the time needed to form a platelet plug in an environment similar to blood vessels. Platelet agonists such as collagen and epinephrine (EPI) are used to activate platelets under high shear conditions. The time taken for the activated platelets to form a plug and occlude an aperture designed to replicate damaged vascular endothelium is measured and reported as the closure time. The instrument cutoff time for nonclosure of the aperture is greater than 300 seconds.
Blood samples were collected into 5 mL blood collection vacutainer tubes containing 3.8% sodium citrate, and samples were kept at room temperature until analysis. An automated hematologic analyzer15 was used to establish a platelet count and hematocrit for each sample before PFA-100® analysis. All samples were analyzed within 2 hours of collection. Samples were mixed well before analysis, and 800 μL of citrated whole blood sample was transferred into a PFA-100® cartridge containing the agonists collagen and EPI (collagen/EPI cartridge),16 and analyzed. Cartridges were stored at 4°C and warmed to room temperature before analysis.
Urinary 11-dTXB2 concentration was measured using a multiplex analyzer17 (which utilizes xMAP technology with fluorescently dyed microspheres to detect low concentrations of substrates) and a commercial competitive enzyme immunoassay kit18 that has been previously validated in the dog. Urine was collected via cystocentesis, batched, and stored at −80°C until analysis. Before analysis, the urine was thawed and the specific gravity was measured. The assay buffer was used to standardize the samples to a urine specific gravity range optimized for the analyzer working range (1.003–1.012). All samples were analyzed in duplicate according to the manufacturer's instructions, and reported in picograms per milliliter of urine. A correction factor was applied to account for the sample dilutions. Briefly, a 96-well plate was prepared by adding 100 μL of the diluted sample to the appropriate well, followed by the addition of 50 μl of both 11-dTXB2 Phycoerythrin Tracer and 11-dTXB2 beads to each well. The plate was placed on an orbital shaker and incubated in the dark at room temperature for 4 hours before analysis.
A biochemistry analyzer19 was used to measure urine creatinine concentration by the Jaffe Reaction. Urine 11-dTXB2 concentration was normalized to the individual's urine creatinine by determining the 11-dTXB2 to creatinine ratio, as previously described.[25, 26]
A single population, repeated measures design was utilized in this study. After visual assessment of Q-Q plots, COX-1 and COX-2 expression, and 11-dTXB2 concentration were deemed to be approximately normally distributed, although the maximal assay closure time was reached at some time points thus affecting normal distribution for the PFA-100 results. Results of platelet COX expression and urine 11-dTBX2 concentration in dogs treated with aspirin were individually assessed by analysis of variance (ANOVA) using the MIXED procedure in SAS for Windows version 9.2.20 Sample time was included in the model as a fixed effect. The repeated measures of dogs over time were accounted for by REPEATED statements assuming an unstructured covariance structure. Differences in least square means with Tukey–Kramer adjustment of P-values were used for multiple comparisons of the three time points for the COX-1 and COX-2 analyses. To accommodate the non-normal distribution of the PFA-100® results, differences in baseline and Day 10 concentrations were evaluated by a Wilcoxon Signed Rank Test using the UNIVARIATE procedure in SAS for Windows version 9.2. A P-value of less than .05 was considered to be significant for all analyses.
Platelet COX-1 expression was measurable in all dogs before drug administration. Time was found to be a significant factor within the model (P = .003). There was a significant increase (P = .003) in COX-1 expression from baseline values by Day 10 of aspirin administration (Fig 1), on average by 250%, with a range of 63–476%. By the end of the washout period, there was a significant decrease (P = 0.002) in COX-1 expression from Day 10 values (Fig 1). There was no significant difference between baseline and washout values (P = 0.140). The aspirin-associated increase in COX-1 expression followed by the decrease after drug discontinuation was observed in all dogs.
Platelet COX-2 expression was also measurable in all dogs before drug administration. Time was found to be a significant factor within the model (P = .031). There was no significant change in COX-2 expression from baseline values after administration of aspirin (P = .124) (Fig 2). By the end of the washout period, however, there was a significant increase (P = .026) in COX-2 expression from Day 10 values. There was no significant difference between baseline and washout values (P = .782). Individual dog responses were variable : COX-2 expression decreased during aspirin administration in 6 dogs (average decrease 70%, range 56–80%), and increased in 2 dogs (by 8 and 116%).
The mean baseline PFA-100® closure time was 167 seconds (range 84–242 seconds). There was a significant (P < .008) increase in closure time by Day 10 of aspirin administration on average by 75%, and mean closure time was 294 seconds (range 274–300 seconds), although the increase was in all likelihood of a greater magnitude because 6 out of 8 dogs attained the maximum reportable closure time of 300 seconds. Increased closure times were seen in all dogs (Fig 3).
The mean baseline urinary 11-dTXB2 to creatinine ratio was 23.7 (range 9.0-42.5). By Day 10 of aspirin administration, there was a significant (P = 0.014) decrease in urinary 11-dTXB2 to creatinine ratio to a mean ratio of 8.5 (range 4.7–12.7), an average decrease of 64%. A decrease in urinary 11-dTXB2 to creatinine ratio was seen in all dogs (Fig 4).
This study demonstrates COX-2 expression in circulating canine platelets. Previously published studies were often unable to detect COX-2 expression in circulating human and canine platelets.[7, 11, 13] As in recently published human platelet studies, addition of a permeabilization step before flow cytometry enabled us to detect COX-2 expression within the platelets of all of the dogs in this study. The functions of the COX-2 isoform in canine platelets are currently unknown.
This study revealed that exposure to anti-inflammatory doses of aspirin altered canine platelet expression of both COX-1 and COX-2. Mature platelets are anucleate and would therefore not be expected to be capable of de novo changes in expression of proteins such as COX-1 or COX-2 in response to endogenous or exogenous stimuli. Nucleated marrow platelet precursor cells such as megakaryocytes contain COX-1 and COX-2,[8, 14] and changes in COX expression that are induced at the level of these platelet precursors would be expected to result in similar changes in platelets derived from affected megakaryocytes. Young newly released platelets also contain residual mRNA derived from precursor megakaryocytes,[27, 28] and retain the intracellular molecular mechanisms necessary for protein production. In fact, in many tissues, COX induction is thought to occur at the post-transcriptional level. Therefore, it is possible that some of the aspirin-induced changes in platelet COX expression that occurred in this study occurred in circulating platelets. Because this study measured COX expression only once during aspirin therapy, 10 days after the drug was commenced, we cannot determine at what time point altered COX expression became apparent in platelets, and thus cannot determine where COX induction occurred. Additional studies evaluating platelet COX expression at several time points during the first week of aspirin therapy, and using flow cytometry to measure COX expression in immature “reticulated” platelets, will help to more precisely determine the sequential effects of aspirin on COX expression at the level of the megakaryocyte, the immature platelet, and the mature circulating platelet.
This study demonstrated a significant increase in platelet COX-1 expression after aspirin administration. While it is generally believed that COX-1 is constitutively expressed in most tissues, up-regulation of COX-1 expression has been demonstrated in human megakaryoblasts during thrombocytopoiesis, although the mechanism of enzyme up-regulation is not fully understood.[29, 30] Up-regulation of COX-1 expression has also been identified in equine jejunal mucosa after periods of ischemia, in human gastric mucosa in association with gastric ulcers, and in humans experiencing rejection of a renal allograft.[31-35] Certainly it appears that in some circumstances, cellular COX-1 expression is induced rather than constitutive, and this study provides supportive evidence that suggests that exposure to aspirin induces COX-1 expression by platelets or platelet precursors in dogs, although the mechanism for this induction remains unknown. Although platelet COX-1 expression was increased, platelet function was concurrently decreased as evidenced by prolonged PFA-100® closure times and decreased urinary 11-dTXB2 concentrations. Prolongation of PFA-100® closure time using the collagen/EPI cartridge is considered to be a sensitive indicator of aspirin-induced platelet dysfunction. Thromboxane B2 is a metabolite of platelet-derived thromboxane A2, and measurement of urinary thromboxane B2 metabolites such as 11-dTXB2 is also considered to be a sensitive indicator of aspirin-induced inhibition of platelet cyclooxygenase activity and resultant decreased thromboxane A2 production.[36-38] Taken together, the increased PFA-100® closure times and decreased urinary 11-dTXB2 concentrations seen in the dogs in this study provide strong supportive evidence of aspirin-induced inhibition of platelet COX activity, a finding that is to be expected given the relatively high doses of aspirin used.
This study also revealed that, in contrast to COX-1, platelet COX-2 expression did not significantly change during the period of aspirin administration. By the end of the post-aspirin washout period, however, there was a significant increase in COX-2 expression compared to Day 10 values, suggesting that aspirin does have an effect on platelet COX-2 expression. Further studies with a larger number of dogs will be needed to determine the true nature of aspirin's effect on canine platelet COX-2 expression.
Previously described flow cytometry techniques for measuring platelet COX-2 have utilized platelet-rich plasma to minimize any potential interference associated with whole blood cellular components. The extra platelet handling associated with the creation of platelet-rich plasma, however, has the potential to activate platelets and thereby artifactually alter test results. Before the initiation of this study, we conducted a preliminary experiment comparing the use of platelet-rich plasma and whole blood for analysis of platelet COX-1 and COX-2 expression via flow cytometry, and found similar results with both methodologies. We therefore elected to use whole blood for this study as the use of whole blood was less time-consuming and not as likely to lead to platelet activation artifacts.
Recently, a COX-2 splice variant was identified from platelet mRNA that appears to be induced by coronary artery bypass grafting in people, with a 200-fold increase in this variant after surgery. The role of this “COX-2a” splice variant in platelet function is unclear, although it does not appear to be involved in thromboxane synthesis. In patients undergoing coronary artery bypass grafting, COX-1 mRNA is the predominant platelet COX isoform, with minimal COX-2 mRNA expression before surgery. Post-operatively, however, there is marked variability in COX expression, with induction of COX-2a mRNA accounting for more than half of the total COX mRNA in some patients. Interestingly, those patients that expressed the COX-2a splice variant were also aspirin resistant after surgery. The human COX-2a splice variant is COX-2 immunoreactive, and it is possible that similar immunoreactive splice variants occur in the dog, and that at least some of the platelet COX-2 detected in this study consisted of splice variants. This study, however, was performed in healthy dogs; therefore it is unlikely that induction of COX-2a splice variants would have occurred. Further studies are needed to identify possible splice variants in dogs and to determine the role of COX-2 and its variants in platelet function.
Some human patients do develop thromboembolic complications despite prophylactic treatment with standard anti-platelet “low-dose” aspirin. This treatment failure has been termed “aspirin resistance.” Although the exact incidence of aspirin resistance in people is undetermined, an estimated 8–45% of patients appear to be poorly responsive to aspirin therapy.[15, 20, 40, 41] The mechanism of aspirin resistance is currently unknown, but it is likely that the etiology is multifactorial. Platelet COX-2, by providing an alternative pathway to COX-1 for platelet thromboxane A2 synthesis, has been proposed as one potentially important cause of human aspirin resistance. Although low-dose aspirin has been shown to produce marked suppression of COX-1 activity in most people, clinical studies have revealed that approximately 20% of patients receiving aspirin have incomplete suppression of urine thromboxane metabolites, suggesting the presence of an alternative pathway (such as COX-2) for thromboxane A2 synthesis. In conditions associated with high rates of platelet turnover and increased thrombopoiesis such as immune-mediated thrombocytopenia and recovery from stem cell transplantation, younger platelets have greater COX-2 expression than mature platelets.[8, 43] One recent study found that while only a small portion (about 10%) of circulating human platelets typically express the COX-2 isoform, up to 60% of all circulating platelets will express COX-2 during conditions of increased platelet turnover and regeneration.[1, 8] Additional studies are required to determine whether or not COX-2 plays a direct role in resistance to low-dose aspirin.
Because all of these dogs responded to relatively high anti-inflammatory doses of aspirin by exhibiting significant platelet dysfunction, as demonstrated by prolonged PFA-100® closure times and decreased urinary thromboxane B2 concentrations, this study revealed no evidence of aspirin resistance. However, because aspirin resistance is typically only considered to be a problem in patients receiving aspirin at lower “anti-platelet” doses, the findings of this study are not surprising. The dose used in this study was much greater than the dose typically used for thromboprophylaxis in dogs. Additional studies are required to determine whether dogs respond in the same manner to anti-platelet doses of aspirin using a typical low-dose regimen of 0.5–1.0 mg/kg PO Q24h, or whether aspirin resistance will become apparent at lower drug doses. However, even at the high aspirin doses used in this study, platelet COX-2 expression in response to drug treatment appeared to vary markedly between individual dogs. Studies designed to evaluate platelet COX-1 and COX-2 expression and to concurrently measure platelet function and thromboxane A2 synthesis in dogs receiving low-dose aspirin may enable us to determine whether or not the variable individual dog COX-2 responses to aspirin at high drug doses correlate with aspirin resistance at lower drug doses.
Aspirin is commonly administered to dogs, both for its anti-inflammatory properties and for its anti-platelet effects in patients at risk for thromboembolic complications. These results demonstrate that circulating canine platelets express both COX-1 and COX-2. Additional research is required to better delineate the functions of platelet-derived COX-2 and its potential role in aspirin responsiveness in both normal and clinically ill dogs.
Aspirin, Major Pharmaceuticals, Livonia, MI
3.8% sodium citrate, Vacutainer tube, Becton Dickinson, Franklin Lakes, NJ
FITC-conjugated monoclonal COX-1, Clone CX111, Cayman Chemical Co, Ann Arbor, MI
FITC-conjugated IgG2b-FITC isotype control, Santa Cruz Biotechnology, Inc, Santa Cruz, CA
Monoclonal anti-human CD9 : RPE, Clone MM2/57, AbD Serotec, Raleigh, NC
Paraformaldehyde, Biolegend Inc, San Diego, CA
Triton X-100, Sigma-Aldrich, St. Louis, MO
FITC-conjugated monoclonal COX-2, Clone CX299, Cayman Chemical Co, Ann Arbor, MI
Monoclonal CD61-purified, Clone JM2E5, Accurate Chemical, Westbury, NY
Goat anti-mouse IgG:RPE, AbD Serotec, Raleigh, NC
FITC-conjugated IgG1 isotype control, Santa Cruz Biotechnology
FACSCalibur, Becton Dickinson, San Jose, CA
CellQuest software, Becton Dickinson, San Jose, CA
PFA-100®, Siemens Healthcare Diagnostics, Deerfield, IL
Abbott Cell-Dyn® 3700, Abbott Laboratories, Abbott Park, IL
PFA Collagen/EPI Test Cartridge, Siemens Healthcare Diagnostics, Duluth, GA
Luminex® 200 System xMAP Technology, Luminex Corporation, Austin, TX
Luminex® 11-dehydro-Thromboxane B2 Kit, Cayman Chemical Co
ACE Alera® Clinical Chemistry System, Alfa Wasserman, Inc, West Caldwell, NJ
SAS for Windows version 9.2, SAS Institute, Cary, NC, 2008