Protein Power: Distortion of the Lipid Bilayer by Rhomboid Proteases

Figure 1. This is an illustration of how the rhomboid proteases distort membranes which do not fit their hydrophobic belt. A typical transmembrane protein is shown in blue, whereas the hydrophobic mismatched rhomboid is shown in red. Note how it changes the shape of the lipids surrounding it and the smaller helices. A rhomboid in a smaller membrane without hydrophobic mismatch is shown in orange. To the left are the three proteins plotted next to the Saffman-Delbrück curve.
Source: Kreutzberger, A. et al. Science 363, 1-9 (2019).

Paper: Rhomboid distorts lipids to break the viscosity-imposed speed limit of membrane diffusion, by Alex J. B. Kreutzberger, Ming Ji, Jesse Aaron, Ljubica Mihaljević, and Siniša Urban

The traditional paradigm for stable, yet dynamic cellular membranes relies on the stability of phospholipids and their strength in numbers. Proteins are imagined merely as working members of this system, either residing on the surface of the membrane or having a hydrophobic transmembrane region capped with two hydrophilic functional caps. In recent years, researchers have discovered a new class of protease which seem to push the limits of this model. Rhomboid proteins, first discovered in Drosophila melanogaster but later found to be ubiquitous in all life1, are a class of intermembrane serine protease. Their function in the cell is to cleave small peptides off other intermembrane proteins, which is the initiating and irreversible step in certain signaling pathways. They use a serine-histidine dyad for catalysis, unlike the Ser-His-Asp triad common in many other enzymes. The human genome encodes five rhomboids, four residing in the plasma membrane, called rhomboid-like 1-4 (RHBDL1-4), and one in mitochondria2.

Kreutzberger et al. investigated the unusual structure of rhomboids and their method of diffusion into the membrane3.  They were previously known to contain six transmembrane α-helices, like normal transmembrane proteins, but they also have an α-helical hairpin protruding out from the core, perpendicular to the other helices1. It is the effect of this and other unique structures that Kreutzberger et al. found distort the membrane in a unique way (Figure 1). Rhomboids have previously been linked to diseases as diverse as Parkinson’s and carcinogenesis3. Thus, it is worthwhile to research rhomboids to further deduce their functions. The biological workaround found by the researchers may be exploited by other proteins, which is important to know because of the possibility of manipulating this property biochemically or biophysically.

Clues leading up to the researchers’ discovery were dots for them to eventually connect. Sanders and Hutchinson found that, when interacting, rhomboids distorted some annular lipids. They speculated that, in membranes, rhomboids distort lipids to make room for cleavage, which occurs in the nonpolar membrane, not exposed to the cytosol. They also mentioned rhomboids could make the adjacent bilayer thinner but didn’t go into detail because they were studying something else4. Bondar et al. ran simulations of membranes that showed nonuniform thinning of the lipids in the presence of rhomboids and their odd protrusions5.

Ramadurai et al. looked at diffusion kinetics of several membrane proteins. They used the Saffman-Delbrück model, which plots the rates of diffusion in and through the plasma membrane. In general, smaller proteins diffuse faster, and this trend makes a smooth curve of molecular radius plotted against diffusion rate. The only deviants from the Saffman-Delbrück model were the two rhomboids they analyzed, and they deviated by a lot. This strongly suggests rhomboids utilize an alternative diffusion pathway6.

Within this growing context of aberrant diffusion, Kreutzberger et al. ran several experiments on three distinct model systems. Cell lines from the human embryonic kidney (HEK239T) and Drosophila (S2R+) were used as biological standards, the former for the possibility of medical applications and the latter because, to date, rhomboids have been studied to the greatest detail in the fruit flies. The third system was a synthetic, in vitro lipid bilayer, used for the more quantitative experiments and compared to observations in vivo.

In all rhomboid experiments, which used RHBDL2, the rhomboid proteases were labeled with HaloTag, which came from a recombinant plasmid in the in vivo systems. This 297-residue peptide was attached to the rhomboid using CRISPR-Cas9 and detected in various ways depending on the experiment. In one experiment, they tethered HaloTag to quantum dots to selectively observe rhomboid diffusion. To make sure the cytoskeleton was not playing a part in rhomboid diffusion, they ran tests with microtubule destabilizing or actin depolymerizing drugs, both with no effect on diffusion rate. To test if diffusion into the membrane was the rate-limiting step in membrane protein cleavage, they used ionomycin precomplexed with magnesium to lower the transition temperature of phospholipids, essentially making the membrane more fluid and increasing diffusion rate. This consequently increased the rate of cleavage, producing kinetic evidence that diffusion is indeed the rate-limiting step in rhomboid cleavage. This is important because it means there could be a large evolutionary pressure for faster diffusion by rhomboid proteases.

The large cytosolic appendage attached to RHBDL2’s N-terminus was speculated to aid with diffusion. In experiments where the cytosolic domain was removed, diffusion rates dropped by 30%, indicating this is the case. Follow-up experiments showed that the appendage repositions the rhomboid in the membrane, further enhancing its ability to distort lipids. Likewise in the in vitro bilayer, replications of Ramadurai’s research were performed. They again found that rhomboids broke the Saffman-Delbrück limit, and even diffused faster than some relatively small peptides.

The most important experiment, however, was the varying of membrane thickness. Rhomboid’s belt, or its intermembrane α-helix region, is thinner than the typical membrane. When in vitro membranes were made thinner by using shorter fatty acid chains, they fit the size of RHBDL2’s belt. Now, since there was no hydrophobic mismatch, the reverse of the previous diffusion rate trend was observed, with normal proteins diffusing faster than rhomboids. This is strong evidence that the advantage rhomboids have is their hydrophobic mismatch with typical membranes, which causes them to be less confined by phospholipids and hence more mobile. This also explains how rhomboids avoid the Saffman-Delbrück limit, because they don’t even fit the conditions of the model in living cells. Other experiments were conducted, but they either assisted the preceding experiments or contributed marginally to the conclusion.

The authors left little doubt that hydrophobic mismatch was the culprit behind the rhomboids’ efficient diffusion in membranes. This inference leaves open the possibility of the natural next step, follow-up research to find other proteins that use the same mechanism for different functions. It could also someday explain the function of other rhomboids which have lost their catalytic ability but remain in cells. These proteins’ ability to distort membranes in a way unlike any other could be used as a mediating factor in other membrane proteins’ function7. In a broader sense, studies like Kreutzberger’s inspire other to think outside the box (or more accurately, to distort the box). Permanently challenging the canonical view of transmembrane proteins may be the most lasting effect of this paper.

Works Cited

  1. Kinch, L. & Grishin, N. Biochim. Biophys. Acta 1828, 2937-2943 (2013).
  2. Urban, S. Cell 167, 1898 (2016).
  3. Kreutzberger, A. et al. Science 363, 1-9 (2019).
  4. Sanders, C. & Hutchinson, J. Curr. Opin. Struct. Bio. 51, 80-91 (2018).
  5. Bondar, A. et al. Structure 17, 395-405 (2009).
  6. Ramadurai, S. et al. Biophys. J. 99, 1482-1489 (2010).
  7. Freeman, M. Annu. Rev. Cell Dev. Biol. 80, 235-254 (2014).

8 thoughts on “Protein Power: Distortion of the Lipid Bilayer by Rhomboid Proteases

  1. The implications that come about with this paper could open a whole new field of research based around similar membrane manipulators. I wouldn’t normally expect trans-membrane proteins to have evolved to have a mismatched hydrophobic section, purposely reducing its ability to fit into the membrane correctly. While this paper certainly showed the effects that these Rhomboid Proteases have on a membrane and how it effected diffusion rates for proteins across that membrane I am now interested in the regulatory systems around these proteases and how various systems have developed ways to utilize this phenomenon.

    1. Investigating regulatory systems would be one good next step. As you may recall, rhomboids are the initiating/irreversible step in cell signaling pathways. As such, they would be highly regulated and their expression would likely depend on the tissue type in the body. RHBDL2 was active in the human kidney. I didn’t find any papers on regulation of rhomboids or their respective genes, but those papers should be in the near future.

  2. I think what was really interesting about this paper is just how quickly the RHBDL2 was able to diffuse, especially because the authors state the protein has 7 transmembrane domains, so it’s not exactly a small protein which would be expected to diffuse quickly. Taking it a step further, they found that the membrane does seem to matter with how quickly diffusion occurs. I think this is an interesting step because I feel like sometimes people think cell membranes are the same because they all have the phospholipid bilayer and there are some proteins throughout, but it makes sense, at least to me, that the surrounding membrane affects the diffusion of proteins. The one thing I question is the very end of the paper. They say non-related proteins also appear to have deformations which may allow for easier diffusion into the membrane, and they claim that a similar mechanism or evolution might have developed which makes sense because a rhomboid shape would make sense for easier diffusion compared to the more traditional shape we think of proteins of being because it is sort of angled and could cut through the membrane, and I wonder how they see this as being beneficial or if it’s useful knowledge for determining the diffusion of other proteins into the membrane. Finally, I wonder what clinical applications this might have or if it is just important to note the importance of the shape and similarity to other proteins

    1. Let me elaborate on the idea of other proteins with the membrane-distorting property. In one of the papers the authors cited, it was mentioned that there are a few proteins in the rhomboid family of proteases (meaning they are rhomboids) which have lost their catalytic function over evolutionary time yet remain in the cell. They have not been removed from the proteome, even though resource efficiency within the cell would normally inactivate the genes that make them; this suggests they still have a non-enzymatic function in the cell. Now that we know they can distort the membrane and diffuse faster, there is the possibility that these non-enzymatic rhomboids could use their faster diffusion rates to influence other biological molecules. One possible mechanism could be that another protein is covalently attached to the rhomboid, which allows the functional protein to perform its intermembrane function faster. This is the type of research the paper I reviewed allows for.
      The authors didn’t specifically state any clinical applications since there was no real translational section. It was more of a pioneering, new chemistry investigation than a practical guidebook. I can’t come up with a clinical application off the top of my head, but maybe there will be one in the future.

  3. From what I can tell the authors suggest that both the shape of the protein and its ability to instigate hydrophobic mismatch is crucial for its ability to rapidly traverse cellular membranes. I am curious as to what your thoughts are on whether or not any particular region of the protein could be slapped onto another through recombinant technology and act as an almost enzyme, to catalyze any protein through the membrane. How conserved are these rhomboid-like motifs between the proteins tested and do these domains work in isolation?

    1. Anything is possible with recombinant DNA.

      No, in all seriousness, I’m sure creating a hybrid protein that carries the best of both worlds from rhomboids and another enzyme class is possible, but it doesn’t happen in real life, and it would be difficult in recombinant organisms because post-translational modification is a thing. The authors said nothing about how conserved the rhomboid structure is or how the unique intermembrane domains work in isolation. If I had to guess, the domains are only somewhat conserved because the primary structure doesn’t matter very much for intermembrane helices, just the secondary structure.

  4. Hi Brian! Great job explaining such an interesting paper! You explained how the most important experiment the authors conducted involved varying membrane thickness in order to observe the rhomboid activity. I think the finding that the thinner membranes fit in the belt of the enzyme is very interesting. I am a little confused as to whether this good fit is due to the mechanistic steps of the enzyme or simply due to the size of the membrane. I wonder if the difference is of any significance, or if there really is a difference at all? Additionally, since rhomboids are known to be linked to diseases and that the hydrophobic mismatch is what allows for the enzymatic efficiency, could future work also involve targeting this characteristic for possible treatments for diseases?

Leave a Reply

Your email address will not be published. Required fields are marked *

css.php search previous next tag category expand menu location phone mail time cart zoom edit close