Data for "Using microfluidics for scalable manufacturing of nanomedicines from bench to GMP: A case study using protein-loaded liposomes."

Nanomedicines are well recognised for their ability to improve therapeutic outcomes. Yet, due to their complexity, nanomedicines are challenging and costly to produce using traditional manufacturing methods. For nanomedicines to be widely exploited, new manufacturing technologies must be adopted to reduce development costs and provide a consistent product. Within this study, we investigate microfluidic manufacture of nanomedicines. Using protein-loaded liposomes as a case study, we manufacture liposomes with tightly defined physico-chemical attributes (size, PDI, protein loading and release) from small-scale (1 mL) through to GMP volume production (200 mL/min). To achieve this, we investigate two different laminar flow microfluidic cartridge designs (based on a staggered herringbone design and a novel toroidal mixer design); for the first time we demonstrate the use of a new microfluidic cartridge design which delivers seamless scale-up production from bench-scale (12 mL/min) through GMP production requirements of over 20 L/h using the same standardised normal operating parameters. We also outline the application of tangential flow filtration for down-stream processing and high product yield. This work confirms that defined liposome products can be manufactured rapidly and reproducibly using a scale-independent production process, thereby de-risking the journey from bench to approved product. Figure 1: Micromixer cartridge designs used within these studies. Schematics illustrate the staggered herringbone micromixer (SHM) with embossed chevrons allowing consistent fluid mixing and the toroidal mixer (TrM) with planar geometry employing centrifugal forces to encourage uniform mixing allowing for greater fluid stream velocities. The flow rate capacities for each of the microfluidic mixers and the microfluidic platforms that are used are listed. Figure 2: Comparison of micromixer design on the physio-chemical attributes of liposomes. Liposomes were prepared using either a staggered herringbone (SHM) in the NanoAssemblr® Benchtop or a toroidal mixer (TrM) in the IgniteTM and their physico-chemical attributes compared. Anionic liposomes (DSPC:Chol:DOPS 10:5:4 w/w) and neutral liposomes (DSPC:Chol 2:1 w/w) were prepared at a 3:1 flow rate ratio and a total flow rate of 15 mL/min and purified by tangential flow filtration. Cationic liposomes DOPE:DOTAP (1:1 w/w) were produced at a flow rate ratio of 1:1 and a total flow rate of 10 mL/min and purified using a 1/10 dilution with Tris to reduce the solvent concentrations. All formulations were prepared at an initial lipid concentration of 4 mg/mL dissolved in ethanol (DOPE:DOTAP) and methanol (DSPC:Chol:DOPS and DSPC:Chol). The liposome z-average diameter (columns) and PDI (open circles) (A) and zeta potential (B) were measured. Results represent mean ± SD of three independent batches. Figure 3: Production of drug loaded liposomes using different microfluidic mixers. Liposomes were prepared using either a staggered herringbone (SHM) in the NanoAssemblr® Benchtop or a toroidal mixer (TrM) in the IgniteTM. Anionic liposomes (DSPC:Chol:DOPS 10:5:4 w/w) and neutral liposomes (DSPC:Chol 2:1 w/w) were prepared at a 3:1 flow rate ratio and a total flow rate of 15 mL/min. All formulations were prepared at an initial lipid concentration of 4 mg/mL dissolved in methanol and loaded with 0.25 mg/mL initial OVA concentration dissolved in PBS pH 7.4. Non-entrapped protein was removed by tangential flow filtration and protein loading was quantified via RP-HPLC. Cationic liposomes (DOPE-DOTAP (1:1 w/w) were prepared at a flow rate ratio of 1:1 and a total flow rate of 10 mL/min. An initial lipid concentration of 4 mg/mL (dissolved in ethanol) and an initial PolyA concentration of 166 µg/mL (dissolved in Tris buffer pH 7.4, 10 mM) was used. Cationic liposomes were purified by dilution and drug loading quantified by using a Ribogreen assay. The liposome z-average diameter (columns) and PDI (open circles) (A), drug loading (B) and zeta potential (C) were measured. Results represent mean ± SD of three independent batches. Figure 4: Production of liposomes entrapping protein are room temperature irrespective of phospholipid transition temperature. Liposomes were prepared from DMPC, DSPC or HSPC in combination with cholesterol at a 2:1 w/w ratio using either a staggered herringbone (SHM) in the NanoAssemblr® Benchtop or a toroidal mixer (TrM) in the IgniteTM. Liposomes were manufactured at a 3:1 flow rate ratio, 15 mL/min total flow rate, an initial lipid concentration of 4 mg/mL (dissolved in methanol) and an initial ovalbumin protein concentration of 0.25 mg/mL (dissolved in PBS pH 7.4) which was quantified by RP-HPLC after purification. All liposomes were prepared at room temperature. Liposomes were purified by tangential flow filtration. The liposome z-average diameter (columns) and PDI (open circles) (A), loading (B), and zeta potential (C) was compared. Results represent mean ± SD of three independent batches. Figure 5: Small-scale production of liposomes using different process parameters. Liposomes (DSPC:Chol 2:1 w/w) were produced by either a staggered herringbone (SHM) in the NanoAssemblr® Benchtop or a toroidal mixer (TrM) in the IgniteTM and the effect of flow rate ratio (A and C) and total flow rate (B and D) investigated on liposome z-average diameter (columns) and PDI (open circles) (A and B) and protein loading (C and D). Liposomes entrapping OVA were manufactured at flow rate ratios of 3:1 or 5:1 and a total flow rates of 12, 15 or 20 mL/min. An initial lipid concentration of 4 mg/mL (dissolved in methanol) and an initial OVA concentration of 0.25 mg/mL (dissolved in PBS) was used. Liposomes were purified by tangential flow filtration and protein loading quantified by RP-HPLC. Results represent mean ± SD of three independent batches. Figure 6: Investigating the impact of solvent choice and lipid concentration when using different microfluidic mixers. Liposomes (DSPC:Chol 2:1 w/w) entrapping OVA were produced using either methanol or ethanol to dissolve the lipid and either a staggered herringbone (SHM) in the NanoAssemblr® Benchtop or a toroidal mixer (TrM) in the IgniteTM. The liposome z-average diameter (columns) and PDI (open circles) (A), protein loading (B; quantified by RP-HPLC) and zeta potential (C) was measured. DSPC:Chol liposomes were manufactured using either methanol or ethanol at a flow rate ratio of 3:1, a total flow rate of 15 mL/min, an initial lipid concentration of 4 mg/mL and OVA concentration of 0.25 mg/mL. Liposomes (DSPC:Chol 2:1 w/w) entrapping OVA were also produced at final lipid concentrations from 0.5 to 10 mg/mL and their particle size (D) and protein loading (E) was measured by a combination of RP-HPLC and micro-BCA. Liposomes were purified by tangential flow filtration and protein loading quantified by HPLC. Results represent mean ± SD of three independent batches. Figure 7: Liposome morphology, lipid recovery and protein release profiles of liposomes produced using different microfluidic mixers. Liposomes (DSPC:Chol 2:1 w/w) entrapping OVA were produced using either a staggered herringbone (SHM) in the NanoAssemblr® Benchtop or a toroidal mixer (TrM) in the IgniteTM. (A) and (B) show the morphology of liposomes produced using the staggered herring bone mixer. (C) and (D) show the morphology of formulations produced using the toroidal mixer. (E) Phospholipid recovery in liposomes produced by each mixer before and after purification via tangential flow filtration. (F) Protein release from liposomes produced by the two different micromixer designs and incubated at 37 °C for 120 h under agitation. DSPC:Chol liposomes were produced at a flow rate ratio of 3:1, and a total flow rate of 15 mL/min. An initial lipid concentration of 4 mg/mL (dissolved in methanol) and an initial OVA concentration of 0.25 mg/mL (dissolved in PBS) was used for (A) to (E). For protein release studies (F), liposome formulations were produced at a 4-folds higher concentration (initial lipid concentration of 16 mg/mL and initial OVA of 1 mg/mL). Protein loading and release was quantified by RP-HPLC. Results represent mean ± SD of three independent batches (E, F). Figure 8: Scale-independent production of liposomes entrapping protein - from bench to GMP. Liposomes (DSPC:Chol 2:1 w/w) entrapping OVA were produced using a toroidal mixer in the IgniteTM, NxGen BlazeTM or a GMP microfluidic manufacturing system. All three systems were run at the same flow rate ratio (3:1). Given the system selected to total flow rate was increased to demonstrate scale-independent production from 12 mL/min (IgniteTM) to 60 mL/min (NxGen BlazeTM) or 200 mL/min (GMP system). The liposome z-average diameter and PDI (A), size intensity (B) and protein loading quantified by micro-BCA (C) are shown. Results represent mean ± SD of three independent batches for the Ignite and NxGen Blaze systems and 1 large-scale batch on the GMP system.

纳米药物（Nanomedicines）因其能够改善治疗效果而广受认可。然而，由于其复杂性，采用传统生产工艺制备纳米药物不仅难度大，且成本高昂。若要实现纳米药物的广泛应用，必须采用新型生产技术以降低研发成本并保证产品的一致性。本研究聚焦于纳米药物的微流控制备工艺。以负载蛋白的脂质体（liposomes）为案例模型，我们实现了从小规模（1 mL）到符合药品生产质量管理规范（GMP, Good Manufacturing Practice）批量生产（200 mL/min）的脂质体制备，其理化属性（粒径、多分散性指数PDI、蛋白载药量与释放行为）均得到严格控制。为达成这一目标，我们研究了两种不同的层流微流控芯片设计：一种基于交错鱼骨形结构，另一种为新型环形混合器设计。本研究首次验证了一款新型微流控芯片的应用，该芯片可在保持标准化运行参数不变的前提下，实现从实验室规模（12 mL/min）到超过20 L/h的GMP级生产的无缝放大。此外，我们还介绍了切向流过滤（tangential flow filtration, TFF）在下游纯化中的应用，可实现高产品收率。本研究证实，采用与规模无关的生产工艺，可以快速且可重复地制备属性明确的脂质体产品，从而降低从实验室研发到获批上市全流程的风险。

图1：本研究使用的微混合器芯片设计。示意图分别展示了带有压印人字形结构、可实现稳定流体混合的交错鱼骨形微混合器（staggered herringbone micromixer, SHM），以及采用平面几何结构、借助离心力促进均匀混合以支持更高流体流速的环形混合器（toroidal mixer, TrM）。文中列出了两款微流控混合器及其适配的微流控平台的流量容量。

图2：微混合器设计对脂质体理化属性的影响对比。分别采用NanoAssemblr®台式系统的交错鱼骨形微混合器（SHM）与IgniteTM平台的环形混合器（TrM）制备脂质体，并对比其理化属性。阴离子脂质体（DSPC:Chol:DOPS 10:5:4 质量比）与中性脂质体（DSPC:Chol 2:1 质量比）均采用3:1的流速比、总流速15 mL/min制备，并通过切向流过滤纯化。阳离子脂质体DOPE:DOTAP（1:1 质量比）采用1:1的流速比、总流速10 mL/min制备，通过用Tris缓冲液以1/10比例稀释以降低溶剂浓度实现纯化。所有制剂的初始脂质浓度均为4 mg/mL：DOPE:DOTAP溶于乙醇，DSPC:Chol:DOPS与DSPC:Chol溶于甲醇。检测指标包括脂质体的Z均粒径（柱状图）与PDI（空心圆点）（A）以及Zeta电位（B）。实验结果为3次独立批次的平均值±标准差。

图3：采用不同微流控混合器制备载药脂质体。分别采用NanoAssemblr®台式系统的交错鱼骨形微混合器（SHM）与IgniteTM平台的环形混合器（TrM）制备脂质体。阴离子脂质体（DSPC:Chol:DOPS 10:5:4 质量比）与中性脂质体（DSPC:Chol 2:1 质量比）采用3:1流速比、总流速15 mL/min制备，初始脂质浓度为4 mg/mL（溶于甲醇），并以0.25 mg/mL的初始卵清蛋白（OVA）浓度溶于pH 7.4的PBS中进行蛋白负载。通过切向流过滤去除未包埋的蛋白，并用反相高效液相色谱（RP-HPLC）定量蛋白载药量。阳离子脂质体DOPE:DOTAP（1:1 质量比）采用1:1流速比、总流速10 mL/min制备，初始脂质浓度为4 mg/mL（溶于乙醇），初始PolyA浓度为166 μg/mL（溶于pH 7.4的10 mM Tris缓冲液）。阳离子脂质体通过稀释法纯化，采用Ribogreen试剂盒定量药物载药量。检测指标包括脂质体的Z均粒径（柱状图）与PDI（空心圆点）（A）、药物载药量（B）以及Zeta电位（C）。结果为3次独立批次的平均值±标准差。

图4：包埋蛋白的脂质体可在室温下制备，不受磷脂相变温度影响。分别采用NanoAssemblr®台式系统的交错鱼骨形微混合器（SHM）与IgniteTM平台的环形混合器（TrM），以2:1质量比将DMPC、DSPC或HSPC与胆固醇复配制备脂质体。制备参数为：流速比3:1，总流速15 mL/min，初始脂质浓度4 mg/mL（溶于甲醇），初始卵清蛋白浓度0.25 mg/mL（溶于pH 7.4的PBS），纯化后通过RP-HPLC定量蛋白含量。所有脂质体均在室温下制备，并通过切向流过滤纯化。检测指标包括Z均粒径（柱状图）与PDI（空心圆点）（A）、载药量（B）以及Zeta电位（C）。结果为3次独立批次的平均值±标准差。

图5：采用不同工艺参数的脂质体小规模制备。分别采用NanoAssemblr®台式系统的交错鱼骨形微混合器（SHM）与IgniteTM平台的环形混合器（TrM）制备脂质体（DSPC:Chol 2:1 质量比），研究流速比（A、C）与总流速（B、D）对脂质体Z均粒径（柱状图）与PDI（空心圆点）（A、B）以及蛋白载药量（C、D）的影响。包埋OVA的脂质体采用3:1或5:1的流速比，总流速分别为12、15或20 mL/min制备，初始脂质浓度为4 mg/mL（溶于甲醇），初始OVA浓度为0.25 mg/mL（溶于PBS）。通过切向流过滤纯化脂质体，并用RP-HPLC定量蛋白载药量。结果为3次独立批次的平均值±标准差。

图6：采用不同微流控混合器时，溶剂选择与脂质浓度的影响。分别采用NanoAssemblr®台式系统的交错鱼骨形微混合器（SHM）与IgniteTM平台的环形混合器（TrM），以甲醇或乙醇溶解脂质，制备包埋OVA的脂质体（DSPC:Chol 2:1 质量比）。检测指标包括Z均粒径（柱状图）与PDI（空心圆点）（A）、蛋白载药量（B，通过RP-HPLC定量）以及Zeta电位（C）。DSPC:Chol脂质体采用3:1流速比、总流速15 mL/min、初始脂质浓度4 mg/mL、OVA浓度0.25 mg/mL制备，溶剂分别为甲醇或乙醇。此外，还制备了终脂质浓度范围为0.5~10 mg/mL的包埋OVA的DSPC:Chol脂质体（2:1 质量比），并结合RP-HPLC与微量BCA法（micro-BCA）检测其粒径（D）与蛋白载药量（E）。所有脂质体均通过切向流过滤纯化，蛋白载药量通过HPLC定量。结果为3次独立批次的平均值±标准差。

图7：采用不同微流控混合器制备的脂质体的形态、脂质回收率与蛋白释放曲线。分别采用NanoAssemblr®台式系统的交错鱼骨形微混合器（SHM）与IgniteTM平台的环形混合器（TrM）制备包埋OVA的DSPC:Chol脂质体（2:1 质量比）。（A）与（B）为交错鱼骨形混合器制备的脂质体的形态；（C）与（D）为环形混合器制备的制剂的形态。（E）为两款混合器制备的脂质体在切向流过滤纯化前后的磷脂回收率。（F）为两种微混合器制备的脂质体在37℃振荡培养120 h后的蛋白释放曲线。上述（A）~（E）的制备参数为：流速比3:1，总流速15 mL/min，初始脂质浓度4 mg/mL（溶于甲醇），初始OVA浓度0.25 mg/mL（溶于PBS）。蛋白释放研究（F）采用的制剂浓度为4倍更高的浓度（初始脂质浓度16 mg/mL，初始OVA浓度1 mg/mL）。蛋白载药量与释放量通过RP-HPLC定量。结果（E、F）为3次独立批次的平均值±标准差。

图8：包埋蛋白的脂质体的规模无关制备——从实验室到GMP生产。分别采用IgniteTM、NxGen BlazeTM平台的环形混合器，以及GMP级微流控制造系统制备包埋OVA的DSPC:Chol脂质体（2:1 质量比）。所有系统均采用3:1的固定流速比。通过提升系统总流速以验证规模无关制备效果，实现从12 mL/min（IgniteTM）到60 mL/min（NxGen BlazeTM）乃至200 mL/min（GMP系统）的放大。检测指标包括脂质体的Z均粒径与PDI（A）、粒径强度分布（B）以及通过微量BCA法定量的蛋白载药量（C）。结果中，Ignite与NxGen Blaze系统为3次独立批次的平均值±标准差，GMP系统为1次大规模批次数据。