Nitrogen Cavitation and Differential Centrifugation Allows for Monitoring the Distribution of Peripheral Membrane Proteins in Cultured Cells

The advantages of nitrogen cavitation over other methods of mechanical disruption are manifold. Perhaps the most significant benefit is its ability to gently yet efficiently homogenize specimens. The physical principles of decompression cools samples instead of generating local heating damage like ultrasonic and friction/shearing based techniques. Cavitation is also extremely efficient at disrupting the plasma membrane. Because nitrogen bubbles are generated within each individual cell upon decompression, the cavitation process is limited less by the cell size, sample size or sample concentration. It therefore offers additional advantages over clearance-based Dounce or needle passage homogenization. Nitrogen cavitation also offers more consistent results as the same disruptive force that is applied uniformly throughout the sample can be reproduced with identical pressures. Moreover, each cell only experiences the disruption process for one single time and subcellular components are, therefore, not constantly exposed to variable disruptive forces. This limits the artefactual fragmentation of organelles. Concern for oxidation of labile cellular components caused by disruption is also mitigated, as nitrogen is used to saturate the cell suspension and is free of supplemental oxygen. The restricted physical and chemical stresses imposed by nitrogen cavitation makes it an ideal technique to study labile enzymes and fragile organelles, and the reproducibility of this homogenization technique is well suited for quantification.

Nitrogen cavitation also offers significant flexibility with regard to sample size, the composition of homogenization buffers, and the degree of organelle integrity. To scale up the homogenization, one simply needs to deploy pressure vessels of larger capacities, without sacrificing the speed, convenience or efficiency of decompression. Any homogenization buffer can be used for nitrogen cavitation; buffer choice can be tailored for compatibility with the requirements of each experiment. One can design a homogenization buffer that matches the intracellular fluid (e.g., high potassium and low sodium) to minimize alterations in organelles. For example, we developed a "relaxation" buffer that minimizes ex vivo actin polymerization and thereby assists in collecting intact cytoplasmic vesicles from granulocytes⁵. The osmotic and ionic features of the buffer can also be modified to control the degree of cell homogenization. Additionally, the aggressiveness of cellular disruption can be controlled by adjusting the nitrogen pressure. Moderate pressure reduces the disruptive force to preserve nuclei, mitochondria and other organelles in an intact state, while high pressure disrupts these organelles. Brock et al. reported success with disruption of Rat Basophilic Leukemia cells while maintaining most nuclei intact at 350 psi with hypotonic buffer¹⁰, which forms the basis of nitrogen pressure used in our current protocol.

Limitations of nitrogen cavitation include the fact that homogenization is uniform throughout the samples and different membrane structures may produce vesicles of similar sizes; this can complicate further separation of the plasma membrane, smooth microsomes, endosomes and Golgi membranes by rate-zonal density gradient centrifugation. Another concern is that the organelles become relatively fragile after homogenization due to the rupture from within. Thus, this method requires careful optimization to prevent organelle breakage in subsequent manipulations. We aimed to address some of these issues in this study.

Separation of cytosolic and membrane fractions by centrifugation has been explored in limited reports detailing the partitioning of protein fractions with a handful of protein markers^15,16. Our method generates a crude separation of soluble and membrane-bound proteins, and does not involve isolation of particular organelles. Yet, it is still a valid concern that the homogenous vesicles of different membranes produced by cavitation may affect the efficiency of separation by ultracentrifugation. We presented a comprehensive survey by immunoblots of the isolation and partitioning of common subcellular organelles after fractionation with our protocol. As expected, purely cytosolic proteins like RhoGDI are exclusively in the soluble fraction, and transmembrane proteins on the plasma membrane (PM) such as EGFR and Na⁺/K⁺ ATPase, were recovered only in the pellet. Transmembrane proteins on organelles such as ER, mitochondria, Golgi and lysosomes were also recovered in the pellet, confirming that integral membrane proteins from different membranes and organelles can be successfully segregated from soluble proteins with our method. This fundamental validation is crucial for the further analysis of peripheral membrane proteins, which is the main purpose of this protocol.

Most of the peripheral membrane proteins exist in both the cytosolic and membrane fractions. Although this is often a true representation of their localization, it may in some instances represent artefactual redistribution of these proteins from the cytoplasmic surface of membranes to the soluble fraction during the homogenization process, since the strength of their association with membranes is not as strong as that of integral membrane proteins. Though nitrogen cavitation minimizes potential redistribution by exposing cells to a one-time, gentle disruption, one should be mindful of such artefacts of biochemical fractionation, and focus on the changes in the partitioning of proteins between cytosolic and membrane fractions across experimental conditions. Nonetheless, the examination of peripheral membrane proteins at various stages of endosomal maturation is possible with this method. EEA1, a Rab5 effector that mediates docking of clathrin-coated vesicles to the early endosomes, localizes to the cytoplasmic surface of early endosomes and was recovered primarily in the pellet fraction. Late-endosome-related Rab7 and Rab9 proteins are also in the pellet fraction. However, there remains an appreciable amount of Rab7/9 in the cytosol. Because RabGDI is able to interact with prenylated Rabs and render these otherwise insoluble proteins cytosolic, Rab proteins tend to be more cytosolic than other endosome-related peripheral membrane proteins.

To characterize the nuclear and organelle integrity with homogenization by nitrogen cavitation, we immunoblotted the isolated fractions for both luminal and cytoplasmic compartments. In this analysis, calregulin is fully released into the soluble fraction, demonstrating that ER is disrupted to form microsomes and thereby leaks luminal content. Given that one half of the total area of membrane in a eukaryotic cell encloses the labyrinthine spaces of the ER, it is unsurprising that its vast network of sac-like or tubular structure fails to remain intact during the expansion of nitrogen bubbles. On the other hand, our analysis of mitochondrial integrity after cavitation shows clear dependence on the cell type, as we observed F₁-ATPase, a protein that is peripherally associated with the inner membrane of mitochondria, in the cytosolic fraction to varying degrees (~35% for HEK-293 and Jurkat, 2% for NIH-3T3). This indicates that nitrogen cavitation at 350 psi with hypotonic buffer does not guarantee mitochondrial integrity. Indeed, others have reported that to preserve mitochondria, cavitation should be performed at pressures lower than 150 psi¹⁷. By contrast, we found that catalase remains in the lumen of peroxisomes. Intriguingly, nuclear proteins are observed in the PNS and pellet fractions. Specifically, HDAC1, a histone deacetylase present throughout the nucleoplasm, is absent in the soluble fraction, indicating that the nuclear integrity is not compromised. This suggests that the nuclear envelope is not ruptured by cavitation, as one would expect the release of HDAC1 into the soluble fraction in that scenario. It remains probable that nuclei are not completely removed during the process of PNS preparation, and the protein remains in the pellet after ultracentrifugation. This likelihood is further supported by the fact that no DNA leakage is detected in the soluble fraction. We conclude, therefore, that using hypotonic homogenization buffer during cavitation at ~350 psi with the addition of Mg²⁺ is able to preserve nuclear integrity in multiple cell lines.

Our results also demonstrate that ribosomes collect entirely in the membrane pellet fraction, suggesting that even free ribosomes in the cytoplasm that are unbound to the ER membranes can be completely sedimented in a non-sucrose-cushion buffer after centrifugation at 350,000 x g for 1 h. Likewise, identification of the autophagy-related protein 12 (Atg12)-Atg5 conjugate as a marker of autophagosomes, reveals strict segregation to the pellet fractions. Its presence in the cytosol may be caused by the fact that the covalent-attachment of Atg12 and Atg5 precedes autophagosomes formation. Cytoskeletal proteins are difficult to assess because of ex vivo polymerization. This phenomenon can be minimized with a relaxation buffer but in the protocol described here, tubulin is likely polymerized ex vivo upon chelation of calcium 18. In our preparations tubulin is found in the pellet suggesting polymerization whereas actin is found in both the P and S fractions.

To characterize potential advantages of nitrogen cavitation in homogenization efficiency, we reviewed some common mechanical disruption techniques. Methods optimized for tissue samples such as blender or mortar/pestle were not evaluated. Sonication was also excluded as it is difficult to disrupt the surface membrane while keeping the internal organelles intact. Our analysis focused on the following methods: freeze-thaw, needle passage and Dounce homogenization, as they are easy to perform and do not require special equipment such as a ball-bearing homogenizer. Jurkat cells were used in this analysis because of their small size, which renders them difficult to homogenize efficiently with traditional liquid shearing techniques that are dependent on clearance, such as needle passage and Dounce. Judged by the concentration of the total protein recovered, we found that cell-size independent disruption methods such as nitrogen cavitation and freeze-thaw indeed show superior efficiency of protein extraction. Similar analysis with cells of larger sizes revealed smaller differences in homogenization efficiencies, though nitrogen cavitation is still superior to alternative physical methods tested (data not shown).

Many cytosolic, endosomal and cytoskeleton proteins were efficiently recovered with all tested methods. In contrast, proteins on the ER and Golgi were optimally recovered using nitrogen cavitation. The slightly lower yield of F₁-ATPase in nitrogen cavitation relative to the other methods may reflect the fact that mitochondria are more likely to remain intact during cavitation and therefore more efficiently removed in the slow speed spin used to generate the PNS. The observation that hexokinase 1, a peripheral membrane protein found both in the cytosol and associated with the outer mitochondrial membrane, behaved in the opposite fashion relative to the F₁-ATPase likely reflects the lability of the association of hexokinase 1 relative to F₁-ATPase, which is sequestered inside the mitochondria. Importantly, although capable of disrupting cells with comparable efficiencies, alternative mechanical disruption methods gave relatively poor recovery of certain peripheral membrane proteins (hexokinase 1 and RAS). Thus, cavitation is our homogenization technique of choice for the purpose of investigating the partition of peripheral membrane proteins. Because Dounce homogenization is widely used to prepare intact nuclei, we used Dounce as a benchmark to evaluate nuclear integrity across other mechanical disruption methods. In our analysis, similar levels of recovery of HDAC1 proteins confirm that nuclei remain intact in Jurkat homogenates prepared by nitrogen cavitation at ~350 psi with hypotonic buffer.

It was apparent that the homogenization buffer used during nitrogen cavitation can significantly affect the disruption efficiency and organelle integrity. While the composition of homogenization buffer should be optimized to the requirements of the intended application, it is worth noting that researchers performing nitrogen decompression of animal tissues use buffer with a low osmotic pressure to extract nuclear contents. Hunter and Commerford showed nuclei swelling and rupture when the solutions were dilute, and nuclei preservation when using isotonic solutions⁷ (which can be achieved by either adding inorganic salts such as sodium chloride or organic solutes such as sucrose or glycerol). The standard isotonic buffer used for isolating subcellular organelles other than nuclei from mammalian cells is 0.25 M sucrose containing 1 mM EDTA and buffered with Tris, HEPES, or Tricine at pH 7.0-7.6. But not all culture cells can be efficiently homogenized in one of the isotonic buffers. Hypotonic buffer (usually 10 mM Tris, HEPES, etc.) is often used to provide necessary osmotic stress to certain cell types to achieve complete homogenization. However, as mentioned before, hypotonic buffers tend to render the nuclei fragile and prone to leakage of DNA when EDTA is included. To promote nuclear integrity, we replace EDTA with KCl and low concentrations of divalent cations such as MgCl₂ or MgSO₄. Mg²⁺ is usually preferred over Ca²⁺ because the latter can activate certain phospholipases and proteases and inhibit RNA polymerase. Once cavitation is finished, EDTA or EGTA can be added back to the homogenates to chelate the cation, if required to inhibit metalloproteases.

Jurkat cells display a relatively large nucleus-to-cytoplasm ratio, making them an ideal cell line to study nuclear integrity. The increase of homogenization efficiency by using hypotonic buffer is minimal when comparing to isotonic buffer supplemented with NaCl. However, cavitation in an isotonic buffer supplemented with sucrose recovers approximately half the proteins extracted by using a hypotonic buffer counterpart. This is unlikely to be a direct result of isotonicity, and the discrepancy between using NaCl and using sucrose to achieve isotonicity requires further investigation. One possible explanation is that salt disrupts electrostatic interactions between proteins whereas sucrose does not. Thus, even at the same tonicity, NaCl has an activity relevant to protein extraction that sucrose lacks. With regard to preserving nuclear integrity, we found increased fibrillarin proteins recovered with isotonic buffers, despite isotonic buffers theoretically being more protective of the nuclei. Curiously, the mitochondrial fraction was more efficiently extracted with isotonic buffer supplemented with NaCl, suggesting that low salt conditions may be suboptimal for mitochondria extraction. Considering cavitation with hypotonic buffer achieves the best balance between homogenization efficiency and nuclear integrity, the hypotonic buffer becomes the buffer of choice in our protocol of cavitation. However, we caution the readers that no buffer is a guarantee of optimal homogenization results. Our protocol should serve as a template for optimization, and pilot testing with desired buffers, cell lines of interest, and other variables (pressure, buffer composition, etc.) must be established for best results.