Foetal bovine serum influence on in vitro extracellular vesicle analyses

Abstract

Extracellular vesicles (EVs) are nanosized lipid bilayer vesicles most notably from either endosomal (i.e., exosomes) or plasma membrane origins (i.e., microvesicles/ectosomes) and released from nearly all mammalian cells (Colombo et al., 2014). An interest in EV research has increased over the past decade, in part due to their participation in complex intercellular communication (Roy et al., 2018). Though EVs are abundant in blood and other biofluids, the investigation of in vitro-derived EVs provides a critical tool for understanding various mechanisms associated with their biogenesis, molecular composition, packaging of specific payloads, and cellular trafficking. Once released, EVs traffic to target cells where they may be taken up to release their payloads via specific mechanisms, and thereby exert their physiological influence (Colombo et al., 2014; Kowal et al., 2014). Although engineered micelles and liposomes have previously been utilized as lipid nanocarriers (Fiandaca & S., 2013; Fiandaca et al., 2011) for many therapeutic applications, EVs have garnered recent interest as drug delivery vehicles (Elsharkasy et al., 2020). Currently, there exist vastly heterogeneous cell culture conditions for EV production and isolation (Consortium, 2017). Therefore, there is a current need to define more standard cell culture conditions for investigating EVs that may accelerate the translation of therapeutic clinical-grade EVs (Lener et al., 2015; Lötvall et al., 2014; Théry et al., 2018). Herein, we present a mini-review on recent investigations reporting the influence of foetal bovine serum (FBS)-supplemented media formulations on cultured cell physiology, EV production/release, and its contaminating presence of vesicular and non-vesicular particles. Additionally, we describe potential solutions and provide recommendations to aid in vitro EV investigators. An international survey observed 83% of International Society for Extracellular Vesicles (ISEV) respondents utilize conditioned cell culture media as their starting material (Gardiner et al., 2016). FBS is a common additive in cell culture and 52% of ISEV respondents utilize serum-containing media for downstream EV analyses, with 59% and 57% of those respondents performing in vitro and in vivo functional studies, respectively (Gardiner et al., 2016). Serum usage, in part due to its ill-defined composition, provides a variety of contaminating particles (e.g., EVs, lipoproteins, and protein complexes, which differ in their physical properties, yet also have similar size, density, and/or RNA components) that confound these investigative results. The growth factors and other constituents within FBS appear to provide a nourishing ecosystem for many cultured cells (Bettger & Mckeehan, 1986). Despite this nourishing milieu, the presence of FBS in culture has raised specific concerns, including the potential introduction of toxins, viral or prion proteins, and mycoplasma, as well as increased culture variability associated with the inconsistency in the FBS manufacturing process (Khodabukus & Baar, 2014; Kirikae et al., 1997; Treadwell, 1963). Moreover, FBS continues to theoretically raise the potential for both xeno-immunization and inadvertent zoonotic agent transmission when considered in clinical applications (Dessels et al., 2016). The major consequence of using native FBS (i.e., untreated FBS that has not undergone any depletion process) to supplement culture media for EV investigations is the requisite introduction of exogenous FBS-derived EVs and other nanoparticles (e.g., protein/growth factor aggregates) within the population of in vitro-derived EVs, thereby contaminating the EV fraction available for downstream isolation (Figure 1). Upon EV isolation, the final fraction will contain a mixture of EVs (and potentially other nanoparticles) derived from both the cultured cells and the conditioned media, thereby confounding any in vitro and in vivo analyses. Moreover, cell-free DNA fragments have been confirmed in FBS and are known to closely associate with FBS-derived EVs on the surface (Shelke, 2018). Unfortunately, current guidelines for FBS manufacturing do not include the routine testing (or removal) of DNA, rendering its presence uncertain within the cell culture system. The stability of the DNA itself, in combination with the stable conformation between the lipid-membrane and DNA fragments may further assist cellular uptake of exogenous DNA, subsequently potentially modulating cultured cell physiology (Langecker et al., 2014). In addition, DNA may be found enclosed within the vesicular lumen of FBS-derived EVs, leaving the possibility for co-isolation with cell-derived EVs (Malkin & Bratman, 2020). Moreover, the characteristics of cell-derived EVs may be affected by the presence of exogenous FBS proteins that may co-aggregate during the EV isolation process. As an example, investigators identified acetylcholinesterase, a proposed marker for small EVs, as a likely non-EV co-aggregate derived from serum, rather than being associated with cell-derived EVs (Liao et al., 2019). Taken together, direct usage of native FBS as a culture media supplement provides major consequences and potential for misinterpretations of EV analyses. Thery et al. (Thery, 2006) proposed the use of either 1) serum-free media; 2) 1% bovine serum albumin instead of whole FBS; or, 3) FBS EV-'depletion' protocols, termed EV-depleted FBS, if the cultured cells require serum supplementation for their growth. To be described throughout this manuscript, it is important to highlight that EV-depleted FBS is not 'EV-free' FBS media as these protocols never 100% deplete FBS-derived EVs. Therefore, we will use the term EV-depleted FBS for when any attempt to deplete FBS-derived EVs has been performed. The gold standard for FBS EV-depletion continues to include diluting FBS media and performing high-speed ultracentrifugation (UC), removing the contained EVs within the pellet, and using the supernatant as the media supplement (Thery, 2006). Of note, performing UC on non-diluted FBS is problematic, since the contained elevated levels of lipids, proteins, and lipoproteins tend to promote aggregation, leaving a less than optimal supernatant for use as a supplement (Thery, 2006). Additionally, during the UC depletion process, free or aggregated growth factors and other proteins may also be removed/reduced due to their similar density as EVs. This removal may also modulate the ability of the FBS to support cell growth (Lehrich et al., 2018). Therefore, it is important to consider this as a potential confounder in experiments comparing EV-depleted FBS versus native FBS. Performing experiments that compare across multiple FBS EV-depletion methods is vital as some methods deplete FBS-derived EVs based on density, while others are based on size. Recently, commercial products are available that are putatively depleted of FBS-derived EVs; however, the exact protocols are not specified (most utilizing polymer precipitants or ultrafiltration), and investigators should utilize these with caution. Since these original FBS EV-depletion protocols were proposed, other researchers have used EV-depleted FBS media in their in vitro investigations. Unfortunately, a growing number of publications have highlighted differing cellular responses to reductions in presumed FBS-derived EV levels in the culture medium through analytical evaluations between cultured cells in native and EV-depleted FBS media (Table 1). Many studies have assessed EV depletion efficiency through reductions in either particle numbers or putative EV-associated RNAs. Size- and concentration-based estimations typically include nanoparticle tracking analysis (NTA) or tunable resistive pulse sensing (TRPS). However, these techniques lack specificity and sensitivity, and are not able to distinguish between EVs and other EV-like nanoparticles (e.g., lipoprotein particles) (Karimi et al., 2018), as NTA may detect concentrations of contaminant low-density lipoproteins (Gardiner et al., 2013). Nanoparticle depletion efficiency is affected by a variety of factors, including UC speed, time, serum dilution, and/or usage of polymer precipitants. Increasing the UC (@120,000 g diluted 1:3) time (e.g., from 2- to 6-h) is known to provide greater nanoparticle depletion (i.e., 7-fold reduction) in the size range of 50–500 nm as evidenced by NTA using a NanoSight NS-500 instrument (Figure 2 A) (Eitan et al., 2015). Additionally, other investigators demonstrated that an 18-h UC (@120,000 g diluted 3:7) removes up to 95% of FBS RNA species compared to only 50% with a 1.5-h UC spin (Figure 2 B, C, D) (Shelke et al., 2014; Wei et al., 2016). In this study, the FBS EV pellet (isolated from EV-depleted FBS) was treated with proteinase K and RNase to exclude other particle-associated RNAs based on the assumption the EV-RNAs are protected within the vesicle. However, the amount of residual EVs in the EV-depleted supernatant was not measured, which makes it difficult to draw definitive conclusions on EV-depletion efficiency (Shelke et al., 2014). Some reports suggest that polymer precipitant methods provide the greatest EV-depletion and reduced variability, while UC methods provide high variability based on each run, batch, and lot differences, and thereby affect final nanoparticle concentrations (Liao et al., 2017). Similarly, quantitative results from our group reported that an 18-h UC (@100,000 g diluted 1:5) resulted in removal of larger (> 250 nm) nanoparticles, while smaller (75–250 nm) nanoparticles remained as measured via NTA with a ZetaVIEW instrument size ranging limits from 50–500 nm (Figure 2 E). Moreover, polymer precipitants, in our hands, resulted in a more heterogeneous mixture of residual nanoparticles (75–500 nm) in the media supplement. Despite both FBS EV-depletion methodologies producing 70% and 75% reductions in nanoparticles, for UC and commercial precipitants, respectively, quantitative analyses indicate significant remaining quantities (109 particles/ml) of nanoparticles within the EV-depleted FBS media conditions (Figure 2 F) (Lehrich et al., 2018). Depending on the exact depletion protocol, various nanoparticles, possibly also including EVs, remain abundant in the EV-depleted FBS media. EVs (1.10-1.19 g/ml) can be separated based on density compared to chylomicrons, very low density lipoprotein (VLDL), and low density lipoprotein (LDL) particles (< 1.063 g/ml), however overlap in density with high density lipoproteins (HDL) (1.063-1.21 g/ml), making their separation from EVs size-dependent (HDL: 4–10 nm) (Brennan et al., 2020). Therefore, since both EV and lipoproteins may be detected by nanoparticle size-based analyses, and both are carriers of exRNAs (Vickers et al., 2011), particle counts and total RNA quantification cannot specifically address EV-depletion from FBS. Instead, FBS EV-depletion efficiency should be determined by quantifying EV-specific protein markers (e.g., CD9, CD63, CD81) via Western Blot (or proteomic assays) in parallel with unconditioned medium controls, including non-depleted FBS, EV-depleted supernatant, and FBS-EV pellet samples. Additionally, amounts of non-EV nanoparticles that overlap in size and density may be determined by quantifying lipoprotein markers (e.g., ApoA-1, ApoB100, ApoB-48, ApoE) in these samples (Brennan et al., 2020; Zhang et al., 2020). Overall, sequential combinations of EV isolation techniques (based on size, density, zeta potential (Zhang et al., 2020), or antibody binding (Mørk et al., 2017)) allow the isolation of nanoparticle populations of interest. For investigations of in vitro-derived EVs, exRNA introduced from FBS should be seriously considered (Figure 3 A). Serum contains a variety of carriers of exRNA including EVs, lipoproteins, and ribonucleoprotein complexes (RNPs) (Tosar et al., 2018) (Figure 3 B). FBS EV-depletion protocols, namely UC, are primarily designed to remove EVs and EV-like particles, leaving uncertainty as to the extent of remaining exRNA carriers present in the media supplement. Such remaining RNA complexes may confound a variety of experimental results, but especially those assessing EV-associated RNA species (Figure 3 C, D) (Wei et al., 2016). One study with RNA-sequencing of EV-depleted FBS media reported that even after a 24-h UC (@100,000 g undiluted), a major proportion of FBS-derived exRNA species remain in solution (Wei et al., 2016). Though contrary to the prior study (Shelke et al., 2014), this may be due to differing spin speeds, dilution factors, and/or RNA quantification techniques. Moreover, this study found that miR-122, miR-451a, which are conserved between humans and cows, are highly abundant in native FBS and remain in the supernatant after EV-depletion protocols (Wei et al., 2016). However, it is not completely understood which RNA types are associated with EVs or with other exRNA carriers, and which exRNA carriers remain in solution after FBS EV-depletion. In fact, it remains difficult to separate individual subclasses of exRNA carriers from plasma or serum (Srinivasan et al., 2019). Argonaute2 complexes are a major reservoir for miRNAs in plasma or serum (Arroyo et al., 2011), and are known to be incompletely removed via UC. Therefore, this class of exRNAs may not be efficiently removed from EV-depleted FBS (Turchinovich et al., 2011), but may be co-isolated with in vitro-derived EVs during polymer-based EV isolation. HDL has been confirmed as a carrier for miRNA, lncRNA, tRNA or rRNA (Allen et al., 2018), and due to their similar density as EVs, exRNAs carried on HDLs may co-precipitate following UC (e.g., density gradient or sucrose cushion). However, the degree of depletion of exRNA carriers achieved in EV-depleted FBS is rarely quantified. It is likely that varying but substantial quantities of exRNA species/carriers (EV-associated or non-EV-associated) remain following EV-depletion protocols. Careful design of EV isolation methods may improve the purity of in vitro-derived EVs and exclude a majority of FBS-derived exRNA carriers (Figure 3 A, E) (Karimi et al., 2018; Mannerström et al., 2019; Onódi et al., 2018). Inclusion of parallel processing controls of non-conditioned FBS-supplemented culture media to compare with the cell-derived EV fraction may be another solution to assess RNA background levels from potential contaminant exRNAs introduced by EV-depleted FBS (Auber et al., 2019; Driedonks et al., 2019). Further, batch-to-batch variations of FBS should be considered when vendors or lots are switched in a laboratory. For additional considerations regarding FBS-derived exRNA contamination and other sources of common laboratory RNA contamination, we refer the reader to the following articles (Das et al., 2019; Murillo et al., 2019; Srinivasan et al., 2019; Tosar et al., 2017). Many experiments suggest that FBS-derived EVs (or EV-like particles) in culture media contribute yet undefined factors important for cultured cell growth and viability. One of the first reports demonstrated that the FBS-derived EV pellet facilitated anchorage-independent growth of breast carcinoma cells (Ochieng et al., 2009). Another group tested a variety of different cell lines (i.e., U87 glioblastoma, HEK-293T, HeLa, SY5Y human neuroblastoma, and N2a mouse neuroblastoma cells) grown in native and EV-depleted FBS media and observed that growth rates and cell viability were substantially reduced in the EV-depleted FBS media for all the cell lines tested, except the U87 cell line. Remarkably, if the FBS-derived EV pellet was 'spiked-in' to the culture media, there is an apparent salvage of growth (Eitan et al., 2015). These negative cell physiological effects associated with EV-depleted FBS media have also been illustrated in primary cell culture systems, including primary human myoblasts (Figure 4 A) (Aswad et al., 2016), primary mouse astrocytes (Figure 4 B) (Lehrich et al., 2018), and cardiac progenitor cells (Angelini et al., 2016). The latter investigation demonstrated that in human cardiosphere-forming cells, FBS-derived EVs appear to modulate cell proliferation, migration, and differentiation. Additionally, cardiosphere structure is affected with differences in sphere volume, overall production, and extracellular matrix generation (Angelini et al., 2016). Lastly, our group revealed that primary mouse astrocytes cultured in EV-depleted FBS media demonstrate suboptimal growth and viability compared to culture in native FBS media (Figure 4 B) (Lehrich et al., 2018). Based on the literature and our own experiences, therefore, the impaired cell growth and viability observed in EV-depleted FBS is likely due to removal of FBS-derived EVs and/or other co-isolated particles. In a series of experiments studying myoblast proliferation, researchers demonstrated that genes important for cell proliferation (i.e., CCND1, SIRT1) were downregulated in EV-depleted FBS media (Aswad et al., 2016). Additionally, FBS-derived EV cargo molecules such as Wnt, TGFß, HSP, sonic hedgehog, SOD, Catalase and survivin may also contribute to these observed cell growth differences (Auber et al., 2019; Eitan et al., 2015). Therefore, researchers are encouraged to properly control for the cell biological influences and their effects on downstream analyses. We suggest that cell proliferation and viability assays be utilized to monitor the effects of cell growth/death, along with the use of an EV potency assay for examining the EVs produced under these 'stressed' physiological conditions providing a preclinical assessment of their therapeutic efficacy, dosing, and biological function (Bobis-Wozowicz et al., 2017; Willis et al., 2017). In addition to impaired cell growth, other investigations have observed induction of specific cellular phenotypes (i.e., alterations in migration, differentiation, inflammation, and secretion) when cultured in EV-depleted FBS media. An airway epithelial model demonstrated that compared to native FBS, EV-depleted FBS media restrained cell migration, which could be salvaged through the direct addition of the isolated FBS-derived EV pellet in a dose-dependent manner (Figure 4 C) (Shelke et al., 2014). A cell differentiation study, utilizing primary cultures of human myoblasts (Aswad et al., 2016), demonstrated that specific genes were differentially expressed when cultured in EV-depleted FBS media. Remarkably, these investigators observed that switching from EV-depleted to native FBS media reversed the induced phenotypic thereby the that FBS EV-depletion protocols modulate cultured cell (Aswad et al., 2016). The of FBS EV-depletion protocols on cell function and has also been & 2015). primary cultured in EV-depleted FBS media release of when with & 2015). Similarly, another observed that cell lines cultured in EV-depleted FBS media increased production, and cell and (Figure 4 D) (Liao et al., 2017). when cultured in EV-depleted FBS media, the increased markers for proteins, and and production (Liao et al., 2017). the reported that may EV production and these results highlight the possibility that conditions present in the EV-depleted FBS media may the characteristics of EV production within cultured cells, and thereby affect downstream analyses. In support of these genes associated with EV and (i.e., were downregulated when cells were cultured in EV-depleted FBS media (Aswad et al., 2016). In these to consider in to the various depletion EV-depleted FBS media may influence cell phenotypes and possibly their and quantitative production of EVs, in vitro EV researchers have utilized serum-free media for EV isolation. However, multiple studies have demonstrated potential when using serum-free media et al., et al., et al., 2014; et al., 2006). Once cells the for EV isolation, switching from native FBS to serum-free media may cellular and et al., in the cellular and potential alterations in EV cargo packaging and release mechanisms et al., 2015). Additionally, there may be of FBS-derived EV and non-EV that the to serum-free media (Auber et al., 2019; Mannerström et al., 2019). serum may cell cells) et al., et al., or the and protein of in vitro-derived EVs et al., et al., 2014). serum concentrations are reduced from to the size total and protein of in vitro-derived EVs were different et al., 2014). Additionally, EV is facilitated through the et al., 2020; et al., 2019; et al., 2018), where specific proteins are expressed during serum et al., 2020). Additionally, cellular introduced by serum-free media may specific associated with EV (i.e., and et al., 2015). Moreover, cellular introduced by different serum concentrations may the and contribute to et al., 2014). In fact, study observed in vitro-derived EVs from cells carried miRNAs and proteins, that growth of breast et al., 2015). study observed in cell and viability under serum depletion in a human cell et al., yet another that cellular may not affect EV size and in human cells et al., cell physiological and EV release need to be considered for cells cultured in serum-free media and tend to differ based on cell types versus primary In addition, serum-free media may not be completely of (Auber et al., 2019). Although serum-free media will not contain FBS-derived EVs, there may exist EV-like particles detected via nanoparticle tracking analysis (e.g., protein in compared to that may with downstream EV analyses et al., 2019). The UC pellet from serum-free media has been to contain protein and vesicular when under transmission along with the presence of on et al., 2019). Moreover, these serum-free media conditions contained RNA which may be derived from other (e.g., et al., 2019). these results suggest that while FBS-derived EVs are in serum-free media, contamination from other sources remain a potential The EV (Consortium, 2017; Lötvall et al., 2014; Théry et al., 2018) and our group (Lehrich et al., has for additional in the reporting of FBS EV-depletion protocols, group has encouraged similar in the of 2018; et al., 2018). A recent study that current in vitro EV isolation protocols may be to clinical based on available methods et al., with another study providing a for manufacturing clinical-grade EV et al., 2018). In this cells are for of specific EV and are in a (e.g., or for EV production et al., 2017). Therefore, cell physiological may not be a primary the of therapeutic EVs is well assessed for and batch-to-batch However, these methods are typically for cell cultures do require serum for growth as and may be in the of primary cell cultures (Lener et al., 2015). A of utilizing EV-depleted FBS or serum-free media is in One group has that (i.e., rather than UC or polymer precipitant is a more FBS EV-depletion and provides an for of cell growth and viability et al., 2018). Additionally, another group that to FBS EV-depletion protocols, such as supernatant removal techniques (e.g., versus or density gradient UC to potentially separate cell-derived EVs from non-EV can substantially affect the efficiency of those methods et al., 2019). These however, have not yet been and along with potential reporting in depletion efficiency due to in nanoparticles et al., et al., 2015; et al., 2014; et al., 2017). Therefore, there remains a need to and media for various cell types to allow more in vitro EV investigations. research has observed that such as human or can be used as a of for cultured cells, to FBS. however, provide their own exogenous EVs et al., 2015; et al., 2017; et al., with the ISEV the use of culture media conditions of and to this contamination et al., 2018). However, in where this is not a culture is with the use of unconditioned medium controls to assess the amount of exogenous that are with EVs of interest. group has using serum-free culture media, with for growth, when in vitro-derived EVs (Lehrich et al., 2018; et al., 2019). there exist to aid in available serum-free culture media there are a of serum supplementation to be for in vitro EV investigations. In addition to by the we suggest additional for reporting in vitro-derived EV the recent of the common of FBS proteins, we for investigators to that the putative proteins identified from isolated in vitro-derived EVs be with this and to provide additional quantitative of et al., 2019). participation will a more of the cell-derived EV Additionally, we support the by and (Auber et al., for the reporting of and and RNA from unconditioned media controls, as a background for in vitro-derived EVs, and for performing to non-vesicular exRNAs (Tosar & 2018). for FBS may be on the experimental The different techniques and protocols for EV proteomic and exRNA isolation and methods have been 2018; 2018). Moreover, factors and process controls (i.e., unconditioned medium controls as a background et al., 2019; Tosar et al., need to be for comparing across differing media controls are especially important in the where some isolation methods with may be more removing nanoparticle populations compared to Lastly, we the to refer to techniques from investigators within the therapeutic viral where more have been to influence of FBS usage in the production, and of therapeutic viral which may be EV analyses. For and are from conditioned serum-free medium et al., where viral release is and from cell and is not This is in with the that for EV therapeutic serum-free culture may be that the of EVs is and are within native FBS culture medium et al., 2019). to FBS-derived contamination multiple of are including density gradient UC, or to the removal of any or et al., et al., 2014). We current should 1) using sequential EV isolation protocols based on size and density (i.e., 2) of the final EV pellet in of size, and protein markers to purity of EVs (i.e., and removal of (i.e., 3) of unconditioned media controls as background such the EV where varying culture protocols are Currently, FBS as a culture media supplement many as for studying in vitro-derived EVs. current FBS EV-depletion protocols lack the ability to the quantities of FBS-derived EVs, exRNA complex and lipoproteins within EV-depleted FBS media, which may downstream cell-derived EV isolation. Additionally, such media with high variability and making analyses if not Based on the the EV may from the use of and media that not only is for cell growth and viability for a variety of cell but also is free of exogenous contaminating FBS-derived EVs and extracellular Although such a media standard is not in the it will isolation of in vitro-derived EV populations that will to clinical The of interest the or methods used in this study or the specified in this reported in this was by the of of the of under to The is the of the and not the of the of applications are to blood as of by and other

Topics & Concepts