Similar but distinct: The impact of biomechanical forces and culture age on the production, cargo loading, and biological efficacy of human megakaryocytic extracellular vesicles for applications in cell and gene therapies

Abstract Megakaryocytic extracellular vesicles (MkEVs) promote the growth and megakaryopoiesis of hematopoietic stem and progenitor cells (HSPCs) largely through endogenous miR‐486‐5p and miR‐22‐3p cargo. Here, we examine the impact of biomechanical force and culture age/differentiation on the formation, properties, and biological efficacy of MkEVs. We applied biomechanical force to Mks using two methods: shake flask cultures and a syringe pump system. Force increased MkEV production in a magnitude‐dependent manner, with similar trends emerging regardless of whether flow cytometry or nanoparticle tracking analysis was used for MkEV counting. Both methods produced MkEVs that were relatively depleted of miR‐486‐5p and miR‐22‐3p cargo. However, while the shake flask‐derived MkEVs were correspondingly less effective in promoting megakaryocytic differentiation of HSPCs, the syringe pump‐derived MkEVs were more effective in doing so, suggesting the presence of unique, unidentified miRNA cargo components. Higher numbers of MkEVs were also produced by “older” Mk cultures, though miRNA cargo levels and MkEV bioactivity were unaffected by culture age. A reduction in MkEV production by Mks derived from late‐differentiating HSPCs was also noted. Taken together, our results demonstrate that biomechanical force has an underappreciated and deeply influential role in MkEV biology, though that role may vary significantly depending on the nature of the force. Given the ubiquity of biomechanical force in vivo and in biomanufacturing, this phenomenon must be grappled with before MkEVs can attain clinical relevance.


| INTRODUCTION
Megakaryocytes (Mks) are large mammalian platelet-producing cells.Under exposure to a cocktail of appropriate cytokines-most notably thrombopoietin (TPO)-hematopoietic stem and progenitor cells (HSPCs) in the bone marrow differentiate to form Mks. [1][2][3] As they mature, Mks become polyploid, accumulating nuclei without dividing, and produce branching extensions called "proplatelets" with which they burrow through the walls of nearby sinusoidal blood vessels in the bone marrow. 1 Subjected to the shear force in these vessels, proplatelets fragment and detach from their parent Mks, forming platelet precursors known as preplatelets. 1,4More recently, research has also demonstrated the migration of whole Mks into the lung vasculature, suggesting a pulmonary origin for some platelets. 5tracellular vesicles (EVs) are submicron-size particles with lipid bilayer membranes that are released by every cell type.7][8] Previously, EVs were categorized as either exosomes (Exos) or microparticles (MPs) based on their biogenesis.MPs are produced directly from the outward budding of the plasma membrane, while Exos originate when exocytosis of multivesicular bodies (MVBs) results in the release of intraluminal vesicles-formed via the inward budding of MVB membranes-into the extracellular space. 6,7Exos, at <100 nm in diameter, are generally smaller than MPs (100-1000 nm in diameter) and have membranes enriched in endosomal proteins.MPs are of special interest for scaled production due to their larger size-which enables the transport of extra cargo-and their relatively simpler isolation.
Nevertheless, the difficulty in separating the EV subtypes by biogenesis method has led the field to embrace the more general "EV" label. 9erefore, we use the abbreviation "MkEVs" to describe the megakaryocytic EVs produced in this study, though our methods (i.e., ultracentrifugation at 25,000g for 30 min.) will concentrate larger particles (formerly MkMPs) and omit many smaller particles (formerly MkExos).
Through the combinatorial effects of their surface receptors, small RNA content, and other cargo, EVs have extensive applications in cell and gene therapies.EVs use various surface receptors to bind specifically to target cells, which subsequently internalize the bound EVs and/or their cargo by way of either endocytosis or membrane fusion. 10,11In many cases, the internalized cargo mediates cellular phenotype. 8,12For instance, MkEVs have been shown to uniquely target HSPCs, with tetraspanins CD54, CD11b, CD18, and CD43 acting as key mediators of MkEV-HSPC binding. 11Subsequent MkEV cargo delivery to the target HSPCs promotes megakaryopoiesis (differentiation into Mks), even in the absence of TPO. 13 Our group has confirmed this function in vivo by using MkEVs to alleviate thrombocytopenia in murine models, 14 a phenomenon which may explain recent reports that MkEV levels are lower in human patients with immune thrombocytopenia. 15Thus, MkEVs offer promise as a treatment for thrombocytopenia and other various Mk/platelet disorders, a finding which takes on particular importance given the severity of current platelet shortages. 16,17Even absent endogenous cargo, MkEVs can be loaded with synthetic cargo and utilized exclusively as delivery vehicles based on their unique capacity for uptake by HSPCs. 18croRNAs (miRNAs) are small, single-stranded RNAs of $22 nucleotides that influence cell phenotype by binding to-and subsequently silencing-messenger RNA (mRNA).Along with protein cargo, miRNA cargo is responsible for the action of EVs on target cells. 19For our system here, the megakaryopoiesis-promoting function of MkEVs can be almost entirely explained by the synergistic action of miR-486-5p and miR-22-3p cargo on HSPCs. 20These miRNAs may mediate JNK and PI3K/Akt/mTOR signaling, with miR-486-5p governing early megakaryopoiesis and miR-22-3p largely responsible for Mk maturation. 20Other work has suggested a role for miR-125b, miR-99a, and/or C-X-C chemokine receptor type 4 (CXCR4), all of which may activate Notch1 via the downregulation of DNA methyltransferases. 21s from a particular source have been generally treated as known quantities with unchanging properties.3][24] Of particular note is biomechanical force, which is ubiquitous in bioprocessing and generally exerts enormous and underappreciated influence over EV quantity and quality. 25We draw parallels with cellular protein glycosylation, which has only recently been appreciated as highly culture-and processdependent. 26r our system here, while the specific binding (tropism) to HSPCs and megakaryopoiesis-promoting function of MkEVs is wellestablished both in vitro and in vivo, nothing is known about the potential variability of MkEV quality that occurs under different culture conditions.Mks are notoriously sensitive to mechanical stimulation, with shear and/or turbulence linked to both increased platelet release 27,28 and faster aging/maturation. 13,29Higher (by up to 47-fold) MkEV yields have been observed following brief (0.5-2 h) exposure of adherent mature Mks to shear stress, 13 but the cargo and efficacy of these MkEV samples were never investigated.Nevertheless, it is increasingly crucial to develop correlations between bioprocessing parameters (such as biomechanical force) and MkEV quantity and quality.Indeed, for all cell types, low yields and unpredictable heterogeneity in EV samples currently pose the greatest impediments to large-scale EV production. 30

| Previous characterization of large MkEVs
We have already extensively characterized the large MkEVs produced from our specific cell culture and isolation protocols. 11,13,14,18,20,31ese findings, which fulfill the requirements outlined by the International Society for Extracellular Vesicles, 9 are outlined in Table S1.Our isolation protocol-illustrated in Figure S1-relies on differential centrifugation, a known "intermediate recovery, intermediate specificity" technique. 9We have also provided new data confirming that MkEVmediated miRNA delivery to HSPCs is a dose-dependent process (Figure S2).In this study, we utilized our prior cell culture and isolation techniques, but individually varied key bioprocessing parametersnamely, biomechanical force and culture age-and subsequently analyzed only the MkEV properties implicated in megakaryopoiesispromoting bioactivity.

| The biomechanical force experiments
As noted in Section 1, the Mk environment in vivo is far from static.
In the sinusoids of the bone marrow, Mks are exposed to shear stress ranging from 1.3 to 4.1 dyn/cm 2 , 4 while the vessel wall shear stress they encounter during migration to the lung vasculature averages 10-15 dyn/cm 2 , 32 with the possibility for a wide range of higher values. 33Given the magnitude of MkEV production required for clinical relevance, shear levels in Mk bioreactors will undoubtedly be higher still.Thus, we imposed two models of biomechanical force on parent Mks: long-term, complex rotational mixing via shake flasks, and brief, defined, and high-intensity shear stress in a novel syringe pump system.
The first model (rotational mixing via shake flasks) is more akin to the complexity of biomechanical force in industrial bioreactors, while the second model (syringe pump system) mimics the average wall shear stress (WSS) observed in the pulmonary vasculature and is easier to scale.For the shake flask experiments, Mks were transferred from T-flasks to shake flasks on day 10 (D10) of the standard 12-day culture and rotated at either 60 or 120 rpm for 2 days.MkEVs were isolated on D12 and subsequently analyzed according to the methods set forth below.Similarly, for the syringe pump experiments, 20 mL of Mks from standard D12 culture were directly loaded into two syringes connected by 250 mm of a 1.58 mm ID silicone tube.The syringes were attached to separate syringe pumps and alternately discharged at a rate of 4448 mL/h for 1.5 h, which corresponded to an average WSS of 15 dyn/cm 2 in the connective tubing, where each Mk spent an average of 2.5% of its time.A Reynolds number of 1000 within the connective tubing confirms the flow is laminar.Viability of the cells subjected to the syringe pump treatment was not significantly different from control cells (Figure S3).We note that because the Mks were suspended in media (i.e., not adherent), they were exposed to a complex range of forces far beyond simple shear, including significant mechanical stress resulting from the rapid expansion/constriction of flow fields at the entrance of each syringe.Note that in all the aforementioned experiments, MkEVs created before the application of biomechanical force were included in the MkEV samples.

| The Mk culture age experiments
Given the established correlation between shear stress and Mk maturation, we hypothesized that Mk age may also impact MkEV production rate, miRNA cargo, and bioactivity.MkEVs from normal (static) Mk cultures were collected after 11, 12, and 13 days and subsequently analyzed per the description later in this section.

| The "recycle" experiments examining HSPC age and passage
Given the large number ($80%) of non-Mks (i.e., CD41 À /CD61 À cells) removed from D7 culture per our established protocols, 11,13 we hypothesized that Mk yield could be increased by "recycling" these non-Mks through our culture process.With this in mind, CD41 À / CD61 À cells removed from Mk (CD41 + /CD61 + ) cultures on D7 were resuspended in D0 media and treated as D0 cells.This process was then repeated a second time for the second "generation" of D7 cultures.Because many or most of the cells in the CD41 À /CD61 À fractions were CD34 + , re-culturing them is akin to "passaging" undifferentiated HSPCs and allows for an examination of the ways in which delayed HSPC differentiation impacts eventual Mk productivity (i.e., MkEV production).

| Long-term, complex biomechanical forces in shake flasks enhance MkEV production in a force-magnitude-dependent manner without affecting MkEV size
A schematic describing the shake flask experiments is present in Figure 1a.MkEVs were counted via both flow cytometry (Figure 1b) and nanoparticle tracking analysis (NTA; Figure 1c) and are plotted on a per-Mk basis.Both methods produced similar trends, though NTAderived counts were roughly two orders of magnitude higher than those calculated using flow cytometry.This phenomenon, reflected elsewhere in the literature, is likely due to an undercount of small (<200 nm) or CD41 À EVs when using traditional flow cytometry and an overcount of non-EV particles (e.g., protein aggregates or lipoproteins) when using NTA. 34Nevertheless, both counting methods suggest that biomechanical force boosts MkEV production, and this trend further appears to be force-magnitude dependent, with progressively-and often significantly-higher MkEV counts observed with each increase in rotational speed (Figure 1b,c).Using NTA to measure MkEV size distribution (Figure 1d-f), we found that there were no significant differences in the mean diameter of the particles (Figure S4).Moreover, flow cytometry was employed to measure the percentages of MkEVs expressing CD54 and CD11b, two known mediators of MkEV-HSPC binding. 11No CD11b expression was noted, and no significant differences were observed in CD54 expression (Figure S4).

| MkEVs produced under long-term, complex biomechanical forces in shake flasks contain lower levels of two key miRNAs involved in stem cell growth and megakaryocytic differentiation
Total and key individual miRNA levels were measured via Qubit fluorimetry and TaqMan RT-qPCR, respectively.PCR-based quantification of miRNA levels was enabled by use of a spike-in control (cel-miR-39-3p) and subsequent application of the 2 ÀΔΔCT method, 35 an increasingly popular technique in the field of EV research. 36,37Levels of each miRNA were normalized to both flow cytometry-and NTAderived MkEV counts.For the shake flask experiments, individual levels of miR-486-5p (Figure 2a,b) and miR-22-3p (Figure 2c,d) were significantly higher in MkEVs from static cultures.As before, the flow cytometry-derived results (Figure 2a,c) and the NTA-derived results (Figure 2b,d) differed by about two orders of magnitude, though trends between samples remained remarkably consistent.Total miRNA levels did not vary significantly between MkEVs, regardless of the MkEV counting method employed (Figure S5), suggesting that the lower individual miRNA levels observed in the MkEVs produced under high rotational speeds do not reflect a dearth of miRNA content generally.Rather, the applied biomechanical forces affect the sorting and loading of these specific miRNAs into the MkEVs.9][40][41][42][43]  2.4 | MkEVs produced under long-term, complex biomechanical forces in shake flasks are less effective in promoting stem cell growth and megakaryocytic differentiation MkEVs produced in the shake flask experiments were co-cultured with HSPCs for 7 days.In each case, a 20:1 MkEV-to-HSPC ratio (as measured via flow cytometry) was used.TPO treatment was used as a positive control.On D7 of co-culture, cells were counted (Figure 3a) and the relative cell fractions expressing CD41 (Figure 3b) and CD42b (Figure 3c) were identified.Cell count is expressed as a fold increase relative to untreated HSPC culture.CD41 is a wellknown marker for Mks generally (including immature Mks), while CD42b is an early marker for Mk maturation. 13The ploidy class distribution (i.e., the fractions of 2N, 4N, and 8N+ cells) for each cell sample was also measured (Figure 3d).Higher ploidy classes are indicative of late-stage Mk maturation. 1,13In general, MkEVs produced under higher levels of shear were relatively less effective, on a per-MkEV basis, at promoting HSPC proliferation (Figure 3a) and differentiation (Figure 3b), and this effect was generally dependent on the magnitude of the force (i.e., cell counts and CD41 expression were higher for samples treated with "static" MkEVs than with "60 rpm" MkEVs, and higher for samples treated with "60 rpm" MkEVs than with "120 rpm" MkEVs).While shear-derived MkEVs were likewise less effective in promoting CD42b expression (Figure 3c), they were similarly effective at promoting high levels of Mk polyploidization (Figure 3d).Differences in Mk polyploidization may not become apparent until well beyond D7; for this reason, polyploidization was not measured in subsequent co-culture experiments.We suggest that for these shake flask experiments, the reduced efficacy of MkEVs produced under shear could be largely ascribed to differential loading of miRNA cargo, particularly given that the fold changes in individual miRNA cargo levels are comparable to the fold changes in MkEV efficacy (Figure 2 vs. Figure 3).Interestingly, the most effective MkEVs promoted Mk maturation better than TPO treatment (as measured by CD42b expression/ploidy class distribution; Figure 3c,d), though TPO still maintained an edge as the most effective means of inducing cell proliferation (Figure 3a), which is consistent with the multifunctional role of TPO in affecting the regulation and proliferation of HSPCs.Cell counts and levels of CD41 and CD42b expression were significantly higher in all MkEV-and TPO-treated cultures than in untreated cultures, as HSPCs do not spontaneously undergo megakaryopoiesis in any significant numbers.Sample flow cytometry data for CD41/ CD42b expression and ploidy distributions are available in Figure 4.

| Brief, high-intensity shear in the syringe pump system increases MkEV production without affecting MkEV size
An experimental schematic for the syringe pump experiments is shown in Figure 5a.Analysis of MkEVs produced from the syringe pump experiments was similar to the analysis performed on MkEVs from the shake flask experiments.Quantities of MkEVs produced under syringe-induced shear were significantly higher than those produced in static cultures, as evidenced by both flow cytometry (Figure 5b) and NTA (Figure 5c) measurements.Here again, the two counting methods differed by roughly two orders of magnitude, with MkEV levels under syringe-induced shear comparable to the levels of shake flask-derived MkEVs, despite a 32-fold reduction in shear exposure time (Figure 5b,c vs. Figure 1b,c).MkEV size distribution and mean diameter following syringe pump treatment are displayed in Figure 5d and do not differ significantly from the mean diameter of control MkEVs produced in static conditions (Figure 1d).No CD11b expression was noted and no significant differences in CD54 expression were observed among the MkEV samples (Figure S6). were significantly lower in the syringe pump MkEVs than in control MkEVs.Also as before, the flow cytometry-derived results (Figure 6a,c) and the NTA-derived results (Figure 6b,d) differed by about two orders of magnitude, though trends between samples again remained consistent.Despite varied quantities of individual miRNA cargo, there was no significant difference between total miRNA levels in syringe pump and control MkEVs, regardless of the MkEV counting method employed (Figure S7).
Total and individual miRNA levels were also quantified on a percell basis for both the shear-exposed and control Mks; the (individual) miR-486-5p and miR-22-3p levels are shown in Figure 6e and Figure 6f, respectively.Notably, there is no significant difference between cellular miRNA levels in either case; combined with the noted differences in the MkEV miRNA levels, these data suggest that miR-486-5p and miR-22-3p cargo is selectively loaded by parent Mks.Indeed, a few quick calculations support this hypothesis.Modeling Mks and MkEVs as spheres and assuming respective diameters of 20 μm and 250 nm (rounded estimates from CellDrop Cell Counter and NTA data), the ratio of Mk volume to MkEV volume is roughly 4 million.Should miRNA cargo be a proportional representation of cellular contents (i.e., "unselectively loaded"), the ratio of cellular miRNA concentration to MkEV miRNA concentration will approach unity.
However, for static culture, this ratio is roughly 400 or 55,000 (flow cytometry or NTA counting) for miR-486-5p and 200 or 33,000 for miR-22-3p; for syringe pump treatment, this ratio is 500 or 77,000 for miR-486-5p and 500 or 72,000 for miR-22-3p.Thus, miR-486-5p and miR-22-3p loading are highly selective in all experimental conditions, with both miRNAs highly upregulated in MkEVs, even under intense shear.We hypothesize that protein chaperones enrich MkEVs with elevated concentrations-relative to parent Mks-of the two miRNAs, and are slightly inhibited by extracellular biomechanical force.

| MkEVs produced under brief, high-intensity
shear in the syringe pump system possess superior capacity to promote megakaryocytic differentiation and are no less effective than control MkEVs in promoting stem cell growth MkEVs produced in the syringe pump experiments were co-cultured with HSPCs at a 20:1 ratio for 7 days (as described previously for the shake flask experiments).Once again, on D7 of co-culture, cells were counted (Figure 7a) and the relative cell fractions expressing CD41 (Figure 7b) and CD42b (Figure 7c) were identified.Sample flow cytometry data for cellular CD41/CD42b expression are shown in Figure 7d-i.Interestingly, although the syringe pump MkEVs possessed less miR-486-5p and miR-22-3p (Figure 6), they were no less effective than control MkEVs in promoting cell growth, and were more than twice as effective in promoting megakaryopoiesis (as measured by cellular expression of CD41, Figure 7b) and Mk maturation (as measured by cellular expression of CD42b, Figure 7c).The mechanisms underlying this phenomenon are unclear, and will require additional investigation (discussed further in Section 3).We suggest that the unique characteristics of the brief, high-intensity shear exerted by the syringe pump imbue the MkEVs with additional miRNA or protein cargo that helps to overcome the reduced efficacy presumably incurred by reductions in the levels of miR-486-5p and miR-22-3p.Indeed, in other cell types, the overwhelming majority of biomechanical force-induced variation in EV bioactivity is directly attributable to changes in miRNA or protein cargo. 25We emphasize that miR-486-5p and miR-22-3p are not the only miRNAs influencing MkEV bioactivity: we have previously observed some MkEV-mediated HSPC growth and differentiation even after inhibiting these key miRNAs, 20 and, as noted in Section 1, others have identified a possible role for other miRNAs. 21

| MkEV production increases with Mk culture age, with the MkEVs retaining consistent miRNA levels and bioactivity
Given the impact of biomechanical force on the acceleration of Mk aging and maturation, we hypothesized that Mk age may affect the MkEV quantity and quality in a manner comparable to the shear stress imposed above.MkEVs from D11, D12, and D13 Mk cultures were isolated and counted using both flow cytometry (Figure 8a) and NTA (Figure 8b).MkEV levels increased dramatically over this timespan, displaying similar trends regardless of measurement technique, despite NTA counts being higher by the usual 2 orders of magnitude.However, MkEVs from the various days did not display significant differences in mean diameter, CD54 expression (Figure S8), total or individual miRNA levels (Figure S9), or bioactivity (i.e., ability to induce growth/megakaryopoiesis of HSPCs during co-culture; Figure S10).Although miR-486-5p levels per MkEV did appear to dip slightly from D11 to D13 when MkEVs were quantified using flow cytometry, this finding was not supported by calculations employing NTA-derived MkEV counts (Figure S9c,d).

| Delayed HSPC differentiation results in Mks with reduced capacity for MkEV production
Given the impact of Mk age on MkEV production, we hypothesized that delayed differentiation of HSPCs (into Mks) may also impact Mk productivity and, therefore, MkEV production.As described previously, CD34 + cells (undifferentiated HSPCs) were continually recultured, such that Mks and their MkEVs could be lumped into three categories: those arising from HSPCs that underwent megakaryopoiesis between D1 and D7 ("D1-D7 differentiation"), those arising from HSPCs that underwent megakaryopoiesis between D8 and D14 ("D8-D14 differentiation"), and those arising from HSPCs that F I G U R E 5 MkEV production rates and size characteristics under brief, defined, and high-intensity biomechanical force.Cells were subjected to 1.5 h of alternating flow in a syringe pump system at the end of D12; MkEVs were subsequently collected and isolated.Mk production for each category was quantified on a per-HSPC basis and plotted in Figure 9c.Similarly, MkEV production for each category was quantified (using flow cytometry) on a per-Mk basis and plotted in Figure 9d.Taken together, the data suggest that as time passes and HSPCs age and replicate, they are progressively less likely to undergo megakaryopoiesis, and the Mks they do produce are increasingly ineffective at producing MkEVs.We also plotted total Mk and MkEV production per initial (thawed) HSPC as a function of time (Figure 9e,f).
Though numeric values were variable due to one highly productive replicate, trends were consistent: both Mk and MkEV production increased substantially following the first "recycle" of CD41 À /CD61 À cells, but not following the second "recycle."From a biomanufacturing perspective, this suggests that attempts to continue deriving Mks/MkEVs from a given HSPC culture after $19 days will prove highly inefficient.

| DISCUSSION
As was once the case with cellular protein glycosylation, EV quality is increasingly recognized as culture-and process-dependent, with numerous-and often overlooked-variables impacting product efficacy.Perhaps the most notable of these variables is biomechanical force, which is ubiquitous in biomanufacturing and therefore vital for the clinical implementation of EV technology.Other variables of note include culture age, media composition and pH, and oxygen tension.Beyond simply expanding the rate of EV production, biomanufacturing processes must ensure that EV quality is standardized.Surface protein expression, miRNA and protein cargo levels, and morphological characteristics including size must all be consistently maintained if EVs are to be successfully harnessed as therapeutics.
This study highlights numerous novel impacts of Mk culture conditions on MkEV quantity and quality.Specifically, relative to static controls, Mks exposed to various types, magnitudes, and durations of biomechanical force produced higher numbers of MkEVs containing lower levels of miR-486-5p and miR-22-3p miRNA cargo.Possibly as a result, some of these MkEVs-those derived under long-term, mild shaking-were relatively less efficacious (in terms of their ability to spur growth and megakaryopoiesis of HSPCs).However, the efficacy of MkEVs derived under the brief, high-intensity shear of the syringe pump matched or exceeded that of control MkEVs.Therefore, where therapeutic applications (e.g., alleviating thrombocytopenia) are desirable, brief, high-intensity shear offers respite from the tradeoffs inherent in long-term, mild shaking, producing MkEVs that are both numerous and efficacious, despite a relative depletion of key miRNA cargo.This article also notes significant increases in MkEV production from D11 to D13, as parent Mks age and mature.However, in contrast with biomechanical force, culture age did not affect MkEV miRNA levels or efficacy.At the same time, Mks arising from older HSPCs that exhibited delayed differentiation were less productive and yielded fewer MkEVs.The key findings of this study are summarized in Figure 10.
Changes in miRNA cargo levels and subsequent EV phenotypes in response to variations in biomechanical force are well-documented for a wide variety of cell types, 25 including endothelial cells, 44,45 fibroblasts, 46 muscle cells, 47,48 bone cells, 49,50 Schwann cells, 51 bronchial epithelial cells, 52,53 and mesenchymal stem cells (MSCs). 54  We distinguish between two different hypotheses for the relative dearth of miR-486-5p and miR-22-3p in MkEVs produced under shear.In the first, miRNA loading machinery is unaffected by shear, but simply cannot keep pace with the increased rate of plasma membrane shedding triggered by the extracellular force.In the second, shear stress triggers signaling cascades that fundamentally alter the biology of the Mks, and, subsequently, their miRNA loading machinery.While the second hypothesis is a feasible explanation for the impact of long-term shear, we question whether short-term shear (e.g., the 1.5 h timeframe employed in the syringe pump experiments) can meaningfully change the EV loading machinery.Moreover, unlike more compact cells, mature Mks are uniquely susceptible to biomechanical forces as a result of their wispy proplatelet extensions, 4 suggesting the first hypothesis may partly explain MkEV production in particular, as MkEVs are derived directly from the plasma membrane.
Under static conditions, however, MkEV production occurs independently of proplatelet formation and is instead reliant on actin depolymerization. 55We have already demonstrated an extreme manifestation of the first hypothesis, wherein Mks are extruded with such force as to rip apart the plasma membrane, creating a multitude of membrane fragments that can spontaneously reassemble to form novel "empty" EVs with a normal landscape of surface proteins. 56In the end, the most likely explanation for shear-induced variation in MkEV cargo is one that includes both hypotheses and encompasses numerous mechanisms, all of which vary with shear duration and magnitude.Indeed, in cultured endothelial cells, shear level has been found to mediate not only the MP production rate, but also the mechanism by which production occurs; 57 a similar phenomenon may occur where MkEV cargo loading is concerned.
The impacts of culture age and delayed HSPC differentiation on MkEV quantity and quality have been hereto unexamined, and the general relationship between cell age and EV quality has been studied only sparingly in other cell types.MSC-derived EVs display notable changes in cargo and reductions in function as their parent cells become senescent. 58Relative to EVs from fresh (D1) platelets, EVs from platelets stored for 5 days are more abundant, enriched in longchain ceramide, and depleted of sphingosine-1-phosphate, inducing lung injury in vivo as a result. 59Among Chinese hamster ovary (CHO) cells, too, EV production rate varies with culture age, 60 and we have recently discovered age-dependent variation in CHO MP cargo, as well (manuscript in preparation).The similarities in MkEV yields between "older" cultures and high-shear cultures are substantial, despite differences in MkEV miRNA cargo and efficacy.Significantly, as noted above, we have previously identified a role for shear stress in promoting early Mk maturation. 13,29The noted similarities between "older" and high-shear cultures may therefore arise partly from this F I G U R E 1 0 Graphical summary of findings.
precise mechanism; that is, the primary effect of shear may be to "age" Mks, prompting them to produce higher numbers of MkEVs and mimicking the observed boost in "late-stage" MkEV production by older cultures.Controversy exists as to the "true" miRNA levels in EVs.Some prior research has identified as little as 1 copy of abundant miRNA species for every 10-100 EVs. 61,62However, another study has noted much higher miRNA concentrations: 12-63 copies of abundant miRNA species in just a single EV. 63One paper of particular note analyzed only large EV fractions (i.e., MPs sorted using flow cytometry) and identified 25-35 copies of abundant miRNA species in each particle. 36A rough approximation of volume in a 100 nm EV also suggests capacity for up to 1000 miRNA-protein complexes. 36While our flow cytometry-based MkEV counts suggest miRNA levels similar to these higher values, our NTA-based MkEV counts reflect much lower miRNA levels.The reality is probably somewhere in the middle.While traditional flow cytometry (used in this study) is ineffective at detecting small (i.e., <200 nm) particles, NTA counts are inflated by non-EV particles such as protein aggregates and lipoproteins. 34Larger EVs are also capable of carrying larger miRNA loads, and our ultracentrifugation protocol enriches for large EVs (i.e., MPs) while eliminating a significant portion of small EVs (i.e., Exos).
While some protein aggregate contamination in EV samples is inevitable-especially for "intermediate specificity" isolation techniques such as differential centrifugation 9 -we suggest that protein aggregate concentration does not vary in this study, and is therefore not a factor in the elevated MkEV bioactivity observed in the syringe pump experiment.While NTA may detect numerous non-EV particles, our flow cytometry protocol detects only CD41 + particles larger than 200 nm, meaning our flow cytometry counts-which are not purely size-basedshould not include non-EV protein aggregates.Therefore, if elevated NTA counts of stress-derived MkEVs resulted from the co-isolation of protein aggregates, we would not expect to see these elevated counts reflected in flow cytometry data.However, fold increases in MkEV production under stress are remarkably similar between the two counting methods, suggesting protein aggregate concentration in MkEV samples is largely constant and thereby independent of biomechanical force.
Our previous work with control MkEVs (produced by D12 static cultures) has also demonstrated that MkEV bioactivity is a direct result of MkEV-HSPC interaction, and relies largely on the delivery of functional miRNA cargo (Table S1). 11,20ture investigations should identify the presence of potentially novel miRNA and protein cargo in MkEVs produced under biomechanical stress.This endeavor will require full miRNA sequencing and proteomic analysis; our group has conducted similar miRNA sequencing of control MkEVs. 20Of particular note are any miRNA(s) and/or protein(s) responsible for the enhanced efficacy of syringe pumpderived MkEVs, as these cargoes will be highly relevant to clinical applications.Once novel miRNA/proteins are identified, each must be individually tested for its ability to exert desirable phenotypes on HSPCs; here again, our previous work provides guidance for future investigations. 20Researchers should also aim to elucidate the particular mechanisms by which miRNAs and proteins are loaded into MkEVs.While the general nature of EV cargo loading is receiving significant new attention and has been well-reviewed elsewhere, [38][39][40][41][42][43] the landscape of mechanisms is diverse, and nothing is known regarding Mks specifically.A thorough understanding of EV loading machinery in Mks would aid in explaining the phenomena characterized in this paper.We have previously discovered that nonapoptotic caspase action mediates the increased release of large MkEVs under very low levels of shear stress. 13Such caspase activity is commonly implicated in the biogenesis of other EVs, 64 and may provide a fruitful starting point for future investigations into MkEV cargo loading specifically.
Figure 11 illustrates common mechanisms by which biomechanical force mediates EV biology, providing additional avenues of inquiry for future MkEV research. 25 investigation into the effects of individual forces (e.g., shear, tension, and/or compression) on MkEV production may also prove valuable.Although these forces are almost never isolated in vivo or among the complex flows of sparged or stirred bioreactors, they nonetheless often affect EV production in different ways, even among similar cell types.For instance, with increasing shear stress, human umbilical vein endothelial cells (HUVECs) produced fewer MPs, 57 while similar human pulmonary artery endothelial cells (HPAECs) subjected to cyclic stretching produced far greater numbers of MPs than their unstretched counterparts. 65A similar phenomenon may be F I G U R E 1 1 Common mechanisms mediating the effects of biomechanical force on EV biology.Biomechanical force is known to increase EV biogenesis and regulate EV cargo in a variety of other cell types.These mechanisms have been reviewed in detail elsewhere 25 and are summarized here as potential avenues of investigation for future MkEV research.

F I G U R E 1
MkEV production rates and size characteristics under complex, long-term biomechanical force.Cells were subjected to rotation in shake flasks at either 60 or 120 rpm during D11 and D12; MkEVs were subsequently collected and isolated.(a) Experimental schematic describing the applied biomechanical force in the context of the overall HSPC/Mk culture process.(b) CD41 + MkEV counts, measured via flow cytometry and expressed on a per-Mk basis.(c) MkEV counts, measured via NTA and expressed on a per-Mk basis.(d-f) Sample NTA size distribution profiles for MkEVs from each of the experimental conditions; error bands represent ±1 standard error of the mean (SEM) of three technical replicates.Mean EV diameters and associated SEM values are calculated from three biological replicates.All other error bars indicate SEM of 3-5 biological replicates.Paired Student's t-tests were performed on all data; *p < 0.05.
Future research must evaluate the presence of potentially novel miRNA cargo in MkEVs produced under biomechanical force (discussed further in Section 3).F I G U R E 2 MkEV miRNA content under complex, long-term biomechanical force.(a) Copies of miR-486-5p per MkEV for flow cytometrybased MkEV counts.(b) Copies of miR-486-5p per MkEV for NTA-based MkEV counts.(c) Copies of miR-22-3p per MkEV for flow cytometrybased MkEV counts.(d) Copies of miR-22-3p per MkEV for NTA-based MkEV counts.Error bars indicate SEM of three biological replicates.Paired Student's t-tests were performed on all data; *p < 0.05, **p < 0.01.

F
I G U R E 3 Bioactivity of MkEVs produced under complex, long-term biomechanical force.MkEVs produced under various levels of biomechanical force were co-cultured with HSPCs at a 20:1 ratio for 7 days.(a) Fold change in cell growth (relative to untreated cells) following co-culture with various MkEV samples.(b) The percentage of cells in each co-culture expressing CD41 (an Mk marker).(c) The percentage of cells in each co-culture expressing CD42b (a marker for early Mk maturation).(d) Ploidy fractions for cells in each co-culture; late Mk maturation is associated with higher ploidy numbers.Error bars indicate SEM of three biological replicates.Paired Student's t-tests were performed on all data; *p < 0.05, **p < 0.01.

2. 6 |
MkEVs produced under brief, high-intensity shear in the syringe pump system are selectively loaded with lower levels of two key miRNAs involved in stem cell growth and megakaryocytic differentiation Total and key individual (i.e., miR-486-5p and miR-22-3p) miRNA levels in the MkEVs produced under syringe-induced shear were measured as before and compared to the miRNA levels in control MkEVs produced in static conditions.As in the shake flask experiments, individual levels of miR-486-5p (Figure 6a,b) and miR-22-3p (Figure 6c,d)

F
I G U R E 4 Flow cytometry data for bioactivity of MkEVs produced under complex, long-term biomechanical force.MkEVs produced under various levels of biomechanical force were co-cultured with HSPCs at a 20:1 ratio for 7 days.Sample flow cytometry data are given for cellular CD41 expression, CD42b expression, and Mk ploidy distribution following co-culture with (a, b) no MkEVs or TPO treatment, (c-e) static MkEVs, (f-h) 60 rpm MkEVs, (i-k) 120 rpm MkEVs, and (l-n) TPO treatment.
(a) Experimental schematic describing the applied biomechanical force in the context of the overall HSPC/Mk culture process.(b) CD41 + MkEV counts, measured via flow cytometry and expressed on a per-Mk basis.(c) MkEV counts, measured via NTA and expressed on a per-Mk basis.(d) Sample NTA size distribution profile for MkEVs produced in the syringe pump; error bands represent ±1 standard error of the mean (SEM) of three technical replicates.Mean EV diameter and associated SEM value is calculated from three biological replicates.All other error bars indicate SEM of 3-5 biological replicates.Unpaired Student's t-tests were performed on all data; *p < 0.05.F I G U R E 6 MkEV miRNA content under brief, defined, and high-intensity biomechanical force.(a) Copies of miR-486-5p per MkEV for flow cytometry-based MkEV counts.(b) Copies of miR-486-5p per MkEV for NTA-based MkEV counts.(c) Copies of miR-22-3p per MkEV for flow cytometry-based MkEV counts.(d) Copies of miR-22-3p per MkEV for NTA-based MkEV counts.(e) Copies of miR-486-5p per Mk following syringe pump-induced shear or control treatment.(f) Copies of miR-22-3p per Mk following syringe pump-induced shear or control treatment.Error bars indicate SEM of three biological replicates.Unpaired (a-d) or paired (e,f) Student's t-tests were performed on all data; *p < 0.05, ns, nonsignificance.

F I G U R E 7
Bioactivity of MkEVs produced under brief, defined, and high-intensity biomechanical force.MkEVs produced in the syringe pump system were co-cultured with HSPCs at a 20:1 ratio for 7 days.(a) Fold change in cell growth (relative to untreated cells) following co-culture with various MkEV samples.(b) The percentage of cells in each co-culture expressing CD41 (an Mk marker).(c) The percentage of cells in each coculture expressing CD42b (a marker for early Mk maturation).(d-f) Sample flow cytometry data for cellular CD41 expression following (d) coculture without MkEVs, (e) co-culture with static MkEVs, and (f) co-culture with sheared MkEVs.(g-i) Sample flow cytometry data for cellular CD42b expression following (g) co-culture without MkEVs, (h) co-culture with static MkEVs, and (i) co-culture with sheared MkEVs.Error bars indicate SEM of three biological replicates.Unpaired Student's t-tests were performed on all data; *p < 0.05, **p < 0.01.underwent megakaryopoiesis between D15 and D21 ("D15-D21 differentiation").An experimental schematic is provided in Figure 9a,b.
However, this article represents the first such documentation of shearinduced miRNA cargo variations in MkEVs.

F
I G U R E 8 Kinetics of MkEV production by mature Mks.MkEVs from D11 to D13 Mks were quantified via (a) flow cytometry and (b) NTA; counts are expressed on a per-Mk basis.Error bars indicate SEM of 3-4 biological replicates, except in the case of NTA-derived D11 MkEV counts (1 biological replicate).Paired Student's t-tests were performed on (a) only; *p < 0.05.F I G U R E 9 MkEV production as a function of HSPC differentiation time.Mks were cultured using standard protocols.Undifferentiated (CD41 À /CD61 À ) cells from each Mk selection were re-cultured.(a) An experimental schematic describing the culture process and the re-culture of undifferentiated cells following each Mk enrichment process.(b) The percentage of cells expressing CD34 in each sample of undifferentiated cells; the colored arrows correspond to the bars on the graph (i.e., the lighter color represents CD34 expression in the first generation of undifferentiated cells, while the darker color represents CD34 expression in the second generation of undifferentiated cells).(c) The average number of Mks produced per CD34 + cell by the original culture, the first generation of undifferentiated cells, and the second generation of undifferentiated cells.(d) The average number of MkEVs produced by Mks from the original culture, the first generation of undifferentiated cells, and the second generation of undifferentiated cells.(e) Cumulative Mks produced per initial HSPC after 26 days.Three biological replicates are represented as three individual trendlines.(f) Cumulative MkEVs produced per initial HSPC after 26 days.Three biological replicates are represented as three individual trendlines; one highly-productive replicate (shown in red) uses a different y-axis.Error bars indicate SEM of three biological replicates.Paired Student's t-tests were performed on a-d; *p < 0.05.
Thus, we hypothesized that MkEV yields, structural characteristics, cargo, and bioactivity may be impacted by biomechanical force and other, related factors such as Mk and HSPC age.Our results demonstrate for the first time that dramatic differentials in MkEV quality occur in response to various magnitudes, durations, and types of biomechanical force on parent Mks.The impacts of Mk and HSPC age, however, are largely confined to variations in