Riboflavin and ultraviolet light treatment potentiates vasodilator-stimulated phosphoprotein Ser-239 phosphorylation in platelet concentrates during storage

The advent of pathogen reduction technologies (PRTs) that can be applied to cellular blood products has heralded a new approach to blood safety for fresh blood component therapy.1,2 These inactivation methods, designed to target pathogen nucleic acid, are based on ultraviolet (UV) light with or without a cross-linking or sensitizing agent.3 The Mirasol PRT system (CaridianBCT Biotechnologies, Lakewood, CO) relies on riboflavin photochemistry as a sensitizer on DNA in white blood cells combined with UV light treat- ment, although much of its pathogen killing effect may occur in the absence of riboflavin.4 It is currently approved in some European countries for treatment of platelet con- centrates (PCs) in plasma or additive solution (AS).5 Exten- sive research has shown that PRT systems can inactivate a broad variety of not only viral and bacterial pathogens in PCs,6 but also the parasites Trypanosoma cruzi7 and Babesia microti.8 However, it is clear that the use of all PRTs involves a trade-off that occurs between the marked increase of the safety profile of the products and the marked decrease of product quality.9 In vitro quality mea- surements of PCs throughout the storage period revealed a contribution of the PRT treatment to the platelet (PLT) storage lesion when compared to the nontreated controls. Alterations include an increase of lactate dehydrogenase release, in P-selectin surface expression, and in a decline in PLT swirling triggered by the PRT procedure.10 Addition- ally, increases of glycolytic flux demonstrated by acceler- ated glucose consumption and mitochondrial activity monitored by an increase of depolarized PLTs have been proposed to result from an increased demand for adenos- ine triphosphate due to an increased release of a-granule contents.11 Finally, in a single-blind crossover study, in vivo recovery and survival of treated PCs were reduced compared to the control units12 as well as shown by reduced corrected count increment.13 Therefore, as these technologies become more refined, it will be essential to minimize product damage. To retain maximum product quality after PRT, it will be necessary to understand the molecular processes that mediated the quality loss of treated and stored PCs. Proteomic technologies have pro- vided a new array of tools to investigate the changes in the PLT proteome over the storage period.14-16

This study applied a semiquantitative proteomic approach to analyze stored apheresis PCs (AP-PCs) and buffy coat PCs (BC-PCs) produced in plasma with and without Mirasol treatment. Gel electrophoresis–based mass spectrometry revealed that the majority of protein changes associated with Mirasol treatment are linked to the cytoskeleton and its regulation of actin dynamics. Focusing on the vasodilator-stimulated phosphoprotein (VASP), biochemical analyses revealed that VASP Ser-239 phosphorylation specifically increased in the Mirasol- irradiated samples compared to the nonirradiated control. This VASP posttranslational modification corre- lated with PLT activation as determined by P-selectin surface expression. This correlation illuminates a novel aspect of the signal transduction pathway triggered by the Mirasol PRT and should be tested in PCs treated by other PRTs.

MATERIALS AND METHODS

Materials

Common chemicals were purchased from Sigma-Aldrich (St Louis, MO) or Fisher Scientific (Ottawa, Ontario, Canada).

Blood PLT preparation and lysis

This study was approved by the research ethics boards of Canadian Blood Services and the University of British Columbia and informed consent was obtained from all healthy volunteers before whole blood donation. Whole blood collection and PLT isolation were carried out by the NetCAD development laboratory of Canadian Blood Ser- vices (Vancouver, British Columbia, Canada). Using stan- dard operating procedures, leukoreduced (5 ¥ 106 per unit) PC preparation from whole blood (donation on Day 0) was carried out 22 hours after collection and controlled cooling to room temperature. The buffy coat fractions were isolated using a semiautomated extractor (Com- pomat G4, Fresenius-Kabi, Bothell, WA), and four buffy coats were pooled for the subsequent preparation of each PLT unit (Day 1). Two BC-PCs (volume of 351.7 ± 17.6 mL and weight of 400.2 ± 18.2 g) with matched ABO blood group were mixed and divided into two equal units in a “pool-and-split” design. Leukoreduced AP-PCs (volume of 258.0 ± 6.6 mL and weight of 307.7 ± 6.8 g) were col- lected on an automated blood collection system (Trima Accel, CaridianBCT), and PCs were divided into two equal units after a 2-hour resting period.

For each PLT preparation, one of these identical units underwent the Mirasol process according to the manufac- turer’s protocol and the other one was kept untreated as control. Briefly, after the addition of 35 mL of riboflavin solution to a final concentration of 50 mmol/L, the unit was exposed to a dose of UV light (6.24 J/mL; 265-375 nm) and rested for 30 minutes before placement in a PLT shaker held at 20 to 24°C. The untreated control bag received 35 mL of a 0.9% saline solution that resembled the same carrier solution used to dissolve the riboflavin ensuring the same starting PLT concentration. All 12 units per group (treated and untreated) for each preparation method of PC (AP-PCs and BC-PCs) were sampled asepti- cally via sample ports into sterile syringes in a laminar- flow hood within 1 hour after treatment as well as after 1, 4, and 6 days of posttreatment storage. It is important to recognize that in this study, buffy coat PLTs are actually treated the day of production, which is the day after dona- tion. Thus while conventional PLT storage studies use ter- minology that reflects the number of days since collection, to facilitate comparisons between apheresis and buffy coat, we have counted the days after treatment rather than after collection. For proteomic analyses, washing and lysis procedures were performed as described previously.17

In vitro PLT quality measurements

For the determination of the mean PLT volume and PLT concentration, the sample was drawn into an ethylenedi- aminetetraacetate tube held for 15 minutes at room tem- perature and measured on a hematology analyzer (ADVIA 120, Siemens, Deerfield, IL). The total PLT count in each container was determined by multiplying the PLT concen- tration from the hematology analyzer by the volume of PLT suspension in the bag.For the determination of the PLT morphology, a fresh PLT sample was mixed with an equal volume of 4% paraformaldehyde solution. Morphology scoring was carried out by a single unblinded technologist who scored slides in a random order on a microscope (Nikon Eclipse 50i, Nikon, Melville, NY) using a modified Kunicki scale in which 100 PLTs are assessed and scored by shape and points assigned: 4 for discoid, 2 for spiny, and 1 for balloon.

Blood gases (pCO2, pO2) and metabolites (glucose, lactate, HCO -, and pH) were determined on a blood gas analyzer (Gem Premier 3000, Instrumentation Laboratory, Lexington, MA). If lactate levels were out of the range of the detector, the samples were diluted with saline. If the pH was out of range, the samples were measured on a portable pH meter (Orion 3 Star pH Benchmark, Thermo Fisher Scientific, Ottawa, Ontario, Canada) and converted to 37°C.
PLT activation was monitored by the expression of P-selectin (CD62P) on the PLT surface using flow cytom- etry. The PLT sample was diluted with phosphate buffered saline to a count of approximately 200 ¥ 109 PLTs/L. A 5-mL sample was then incubated for 30 minutes with both CD62-phycoerythrin and CD42a-fluorescein isothiocyan- ate (FITC) antibodies (Beckman-Coulter, Mississauga, Ontario, Canada). The PLTs were analyzed on a flow cytometer (FACS Canto II, BD Biosciences, Mississauga, Ontario, Canada); PLT activation was determined as the percentage of the CD62 positive events within a gate set by the pan-PLT marker CD42a (Beckman-Coulter).

Potential signs of apoptosis were monitored by changes in the exposure of phosphatidyserine on the PLT surface using annexin-V binding. The PLT sample was diluted with phosphate buffered saline to a count of 100 ¥ 109 PLTs/L, incubated for 30 minutes with annexin-V FITC (BD Biosciences) in a calcium-containing buffer and analyzed by flow cytometry.

All PLT units were tested for sterility after the Day 6 sampling by inoculation of 8 to 10 mL from each unit into BPA (aerobic) and BPN (anaerobic) culture bottles (BacT/ ALERT, bioMérieux, Marcy l’Etoile, France) followed by incubation in an incubator (BacT/ALERT, bioMérieux) for 6 days. None of the units tested yielded results positive for bacterial growth.

Proteomic analyses

Large-format gel electrophoresis (30 ¥ 30 cm) was used to separate 50 mg of PLT lysate, and gels were subsequently stained with silver or Coomassie blue R-250 (Bio-Rad Laboratories, Mississauga, Ontario, Canada). Protein pro- files were generated using a program on a image acquisi- tion system (GeneSnap and Chemigenius, respectively, Syngene, Frederick, MD). Proteins bands that showed reproducible changes due to Mirasol treatment or storage were subjected to spot excision, in-gel digestion and sequence analysis by tandem mass spectrometry (MS/MS) using a LC-MS/MS system (Q-Star, Applied BioSys- tems, Foster City, CA). Extracted peptides were identified by MASCOT searches18 against current Swiss-prot data- bases using standard acceptance criteria (ion scores above significance threshold [p  0.05], with two or more peptides matching).

Immunoblot analyses

PLT lysates were separated on sodium dodecyl sulfate– polyacrylamide gel electrophoresis gels and blotted onto nitrocellulose membranes (Bio-Rad Laboratories). The membrane was probed with primary antibodies against 14-3-3 (Abcam, Cambridge, MA), cofilin, VASP, phospho-VASP (Cell Signaling, Danvers, MA), or b-actin (Sigma-Aldrich), followed by their respective secondary antibodies (Licor, Lincoln, NE). Protein band intensities were determined by densitometry using the imaging analysis software on a bioimaging system (Odyssey and LICOR, respectively, Licor).

Statistical analyses

Means and standard deviations (SDs) were calculated for all variables. A two-way analysis of variance (ANOVA) with repeated measures followed by post hoc testing to identify the source of significance was performed to deduce significant interactions between units with and without riboflavin and UV light treatment and a p value of less than 0.05 was considered to be significant. For correla- tion studies, correlation coefficients were calculated and critical values were used to determine significance.

RESULTS

In vitro analyses of PCs

For most variables no differences were detectable imme- diately after Mirasol treatment compared to the untreated unit except pO2 and PLT activation (Tables 1 and 2). During storage, the PLT count and the mean PLT volume did not change significantly (p  0.05) for either PLT product. Morphology assessed by Kunicki scoring showed a significant effect (p  0.05) on PLT integrity for both PLT products independent of PRT treatment during storage. However, after 6 days of storage the PLT morphology showed no significant difference (p  0.05) between the treated and untreated samples for both PLT preparation methods. The partial pressure of O2 was reduced by approximately two-thirds in both PLT products. The partial pressure for CO2 changed approximately 50% during storage, independent of the Mirasol treatment for both PLT products; the partial pressure for O2 was very low immediately after Mirasol treatment compared to the untreated sample. Although the difference between the irradiated and nonirradiated samples decreased during storage, the oxygen tension was still significantly different (p  0.05) at the end of storage. Interestingly, for AP-PCs a difference of 7.8% PLT activation caused by the treatment was observed compared to the BC-PCs, which exhibited only a 1.3% difference. Furthermore, statistical analyses using repeated two-way ANOVA revealed a significant dif- ference (p  0.05) in the PLT activation assessed by CD62P measurement comparing both PLT products in the irradi- ated arm, which was however not significant (p  0.05) for the control arm.

PLT metabolism was strongly affected by the treat- ment in both types of PLT products showing a significant accelerated glucose consumption and subsequent lactate accumulation in the Mirasol irradiated units (p  0.05). The nontreated samples showed the commonly observed 25% decrease of glucose, one of many changes temporally associated with the development of the storage lesion in untreated PLTs; however, Mirasol treatment resulted in a 45 and 80% decrease in glucose consumption for the BC-PCs and AP-PCs, respectively. The pH values were sig- nificantly different in the treated and untreated samples (p  0.05) throughout the 6-day storage for both PLT prod- ucts. The difference of metabolite levels observed between BC-PCs and AP-PCs was mirrored in their respective pH values that showed a smaller difference for the BC-PCs between the Mirasol irradiated and nonirradiated samples compared to AP-PCs. Consequently, the treated AP-PC sample after 6 days of storage was below the AABB stan- dard of pH 6.2.19 The observation that BC-PCs seemed to maintain the pH level better than AP-PCs is reflected in the bicarbonate (HCO -) concentration.

For PLT activation, the Mirasol treatment contributed to the PLT storage lesion–derived effect the same way for both PLT products; although AP-PCs appeared to have a higher overall activation due to a higher basal activation. Annexin V binding showed the same trend for both BP-PCs and AP-PCs with a significant increase of phos- phatidylserine exposure (p  0.05) triggered by Mirasol treatment throughout the 6-day storage period compared to the nonirradiated units.

Proteomic analysis of PLT samples treated with and without Mirasol technology

To gain further understanding about molecular mecha- nisms triggered by PRTs, we applied a semiquantitative proteomic approach to assess the alterations of the PLT proteome caused by Mirasol treatment. A large-format gel electrophoresis was selected due to its ability to display and compare all the samples collected for the entire storage period of both types of PLT preparations at the same time (Fig. 1). The similar protein profiling of AP-PCs and BC-PCs showed only small differences that may be due to donor-donor variation or the nature of the two different methods of preparation. To identify protein dif- ferences between Mirasol-treated and untreated samples as well as samples collected immediately after Mirasol treatment (production and treatment [P/T]) and after 6 days of storage (Day 6) we generated a protein profile by scanning the band intensity pattern from the top to the bottom of the respective lane for BC-PCs (Fig. 1). There were no significant differences detectable between before and shortly after treatment (data not shown). However, increasingly pronounced alterations could be observed during storage when comparing units sampled on the day of P/T and on Day 6 after Mirasol treatment (Fig. 1). With respect to specific protein changes, only protein changes that were reproducible in four of six individual sample sets were considered for further analyses. These proteins (Fig. 1, #1 to #14) were excised from the gel and subjected to in-gel digestions and mass spectrometric analyses. The protein identifications are summarized in Table 3 along with the calculated change in protein concentration determined by the individual protein peak by comparing the protein amount on Day 6 of storage versus the day of P/T in the Mirasol-irradiated sample as well as the (poten- tial) protein function. As seen in Table 3 most proteins identified in our Mirasol study are involved in interactions with the actin cytoskeleton, suggesting a link to actin dynamics. Furthermore, although the current PRTs for PCs are based on different methods, we have compared our proteomic results to a recent proteomic study that analyzed proteins changed by using a UV-C–based PRT, UV-B, and gamma irradiation.20

Validation of proteomic findings by immunoblot analyses

Proteomics identifies most of the proteins in an individual protein band; therefore, it is important to validate the results obtained from the mass spectrometric analyses. Immunoblot analyses are the most common validation tool and representative sets for 14-3-3 and cofilin are shown in Fig. 2. These analyses demonstrated a decrease of the protein concentration of 14-3-3 in AP-PCs and BC-PCs of approximately 40% during the 6-day storage in both the irradiated and the nonirradiated sample (Fig. 2), which is in agreement with the calculation from the pro- teomic result (Fig. 1 and Table 3). As a second example, a 1.5-fold increase of the concentration of cofilin in both AP-PCs and BC-PCs during the 6-day storage in both the Mirasol-irradiated and nonirradiated sample could be detected in the immunoblots (Fig. 2), which was also in agreement with the proteomic data (Fig. 1 and Table 3).

Phosphorylation levels of VASP

Since the majority of proteins found to be changing in our semiquantitative proteomic approach are involved in the structure and regulation of the cytoskeleton, we hypoth- esized that Mirasol treatment affects actin dynamics. Besides cofilin, which is described above, the VASP demonstrated a slightly increased protein concentration during storage (Fig. 3A), which is in agreement with the calculation from the proteomic result (Table 3). Activation of VASP is regulated by the phosphorylation of two main residues, Ser-157 and Ser-239. Therefore, we monitored the levels of these individual phosphorylated protein iso- forms by immunoblot (Figs. 3B and 3C). For AP-PCs, an increase in VASP phosphorylation at Ser-157 was observed starting on Day 2 of storage while for the BC-PCs the phos- phorylation level had increased after Day 4 of storage; however, this increase of this posttranslational modifica- tion seemed to be independent of the Mirasol treat- ment for both PLT products. Phosphorylation on Ser-239 showed a continuous increase until Day 6 of storage for BC-PCs, in contrast to AP-PCs, which not only displayed a more pronounced increase in the phosphorylation level, but showed a striking effect from the Mirasol treatment throughout the 6-day storage.

Correlation of VASP-P Ser-239 with PLT activation Since VASP is involved in regulating cytoskeletal rear- rangement and this process is linked to microtubule trafficking and vesicle secretion, we evaluated whether there was a potential relationship between the degree of phos- phorylation of VASP on Ser-239 and PLT activation, which was determined by the expression of P-selectin on the PLT surface due to the secretion of a-granule. As demon- strated in Fig. 4, we found a strong exponential correlation between these two variables for the AP-PCs (r2 = 0.922; p  0.01) and a linear correlation for BC-PCs (r2 = 0.904; p  0.01), suggesting a linkage between the Mirasol- triggered VASP Ser-239 phosphorylation and PLT activa- tion caused by the Mirasol treatment. When a linear trend line was applied to the AP-PC results, a markedly reduced r2 value was obtained. To explore the meaning of the observed correlation between VASP Ser-239 phospho- rylation and PLT activation by P-selectin expression, we calculated the correlation coefficients between the phos- phorylation of VASP and pH, morphology, or annexin V binding. No relationship showed a significant correlation between either PLT product and any of these assays (data not shown), suggesting that there is a biologically relevant correlation only with P-selectin expression; all r values were below the critical value of 0.576 required to achieve significance at the 0.05 level.

DISCUSSION

Several studies have been conducted to assess the impact of the Mirasol technology on PLT quality.21-23 Monitoring in vitro and in vivo variables revealed that the Mirasol treatment caused a significant loss of in vitro PLT quality reflected as increased PLT activation and metabolic activ- ity as well as reduced recovery and survival in vivo.10,12,24 The majority of Mirasol studies were carried out using AP-PCs and mainly focused on comparing the influence of PLT additive solutions on in vitro quality.21 BC-PCs have only been described as part of one comparative study with apheresis PCs both prepared in plasma in a similar work flow to that used in our study.25 Our results are largely in agreement with this investigation both in terms of vari- ables that do not change such as PLT concentration and PLT (mean) volume and in terms of the trend of changes including pO2, pCO2, and morphology. The values for glucose, lactate, PLT activation determined by P-selectin surface expression, and phosphatidylserine exposure measured by annexin V binding suggest a strong similarity between Mirasol treatment and the features of PLT storage lesion development. Except for pH, all in vitro variables analyzed in this study showed no significant difference between the two types of PLT products. Although no dif- ference was observed in the lactate production in the two arms of the study, Mirasol-irradiated versus nonirradi- ated, the pH decreased significantly by Day 5 and addi- tionally on Day 7 in the AP-PCs, a finding not seen in the BC-PCs. Consequently, some of the units at Day 7 in the Mirasol-irradiated study arm of the AP-PCs would not have passed the criteria set by AABB standards (pH ≥ 6.2). This difference in the pH between the PLT products was not observed in the study by Li and colleagues25 and could were not placed in context with the functional data because a biologic explanation is lacking at this point of time.
In this study, we used large format one-dimensional gel electrophoresis, which allows comparative analyses of the entire set of samples before and after treatment as well as during subsequent storage. A fairly small number of differences in individual protein concentrations could be detected, some of which were also found to be changing in the UV-C proteomic study20 (Table 3). However, due to the different nature of the irradiation technology, to the pro- teomic analytical methods,16 or to the timing of sample collection, no conclusion in terms of similar molecular mechanisms in PLTs can be drawn. Furthermore, the most common early signs are alterations in posttranslational modification of proteins that are difficult to detect by the setup of our proteomic approach.

Alterations observed over the storage period in the samples from the untreated study arm reflect the changes observed in earlier studies analyzing protein changes in the PLT proteome during blood bank storage.15,16 Impor- tantly, they provide the base level for changes during storage and must be dissected from the changes caused by the Mirasol treatment alone. Therefore, the main focus of our analyses was on the alterations of the proteome profile during storage induced by Mirasol treatment. We found 14 reproducible protein changes that are highlighted in Fig. 1B and summarized in Table 3; the changes of several of these proteins were confirmed by immunoblot analyses.

Proteins changing due to Mirasol treatment were mainly cytoskeletal proteins involved in the maintenance of the actin structure and in the regulation of its dynamics (Fig. 2). Thus, we focused on regulator proteins of actin dynamics. The VASP was an interesting candidate because it is a key player in actin cytoskeleton rearrangement and controls the dynamic actin turnover at the barbed ends of actin filaments. Activation is achieved by phosphoryla- tion of the major sites Ser-157 and Ser-239 triggered by cAMP- and cGMP-dependent protein kinases, respec- tively. Detailed biochemical analyses revealed that Ser-157 controls subcellular VASP distribution, whereas phospho- rylation at Ser-239 impairs F-actin assembly and accu- mulation.33 Furthermore, VASP Ser-157 phosphorylation appeared to be involved in the negative regulation of the integrin aIIbb3. In light of our findings, the increase in VASP Ser-157 throughout storage and accelerated by Mirasol treatment could explain the impaired aggregation reported after Mirasol treatment.10,22 Since actin rear- rangement is involved in granule secretion34 we sought to test for a relationship between the VASP Ser-239 phospho- rylation and the PLT activation as determined by CD62P expression after a-granule secretion. Strikingly, data cor- relations suggest a link between VASP and granule trans- port for both PLT products (AP-PCs and BC-PCs); however, this observation does not confirm causality.

Our findings suggest a novel molecular mechanism triggered by the application of the Mirasol technology to PCs linking the signal transduction caused by irradiation to PLT activation via integrin activation and degranulation mediated by actin cytoskeleton rearrangement. Possibly, the phosphorylation level of VASP on Ser-239 may be a useful protein marker to monitor PLTs in vitro while refin- ing the procedures of the Mirasol technology. Interest- ingly, VASP phosphorylation has been used to evaluate the individual response to a clopidogrel loading dose before percutaneous coronary intervention and predict postpro- cedural major adverse cardiovascular events.35 While our studies were restricted to Mirasol, there are some techni- cal similarities among the three major PLT PRT methods. It remains to be determined if the PLT damage that is the trade-off for increased protection from pathogens using these TEPP-46 technologies also shares a common root cause.