Hacker News new | comments | show | ask | jobs | submit login
A Molecular View of HIV Therapy [video] (rcsb.org)
54 points by cvarjas on Mar 21, 2017 | hide | past | web | favorite | 6 comments

Pretty cool visualization.

HIV integrase is used a lot in many gene therapies. It is very good at 'inserting a genetic payload' into a human's genome. So we often co-opt it to insert desirable sequences (see sickle-cell disease discussion [1]). Everyone's favorite Cas9 is able to cut DNA at particular sequences, but is not of much help actually getting a payload to load inline at the location that it homes in on. The HIV integrase is about the opposite, it's great at getting a payload to load inline, but it inserts relatively randomly (which is dangerous - an insertion at random spots could put code in the middle of an important oncogene). Ideally we'd have a kind of hybrid integrase/Cas9 that is able to both target, and insert a payload dna inline into the genome - at a specific site only.

The therapies described in the video are all small molecules. Some of the more interesting, newer therapies however are instead protein mimics that actually play a higher-level role in 'deceiving' rather that just trying to 'jam' the virus (see the eCD4-IG synthetic protein [2]).

[1] https://news.ycombinator.com/item?id=13781549

[2] https://serotiny.bio/notes/proteins/ecd4ig/

I'm curious, have you heard any talk about using a CRISPR/Integrase combination to insert extra RB gene(s)? I feel like inserting extra copies of a tumor supressor gene would really help decrease some cancers, especially those like RB that involve the two hit hypothesis.

My background is not clinical, so I'm not familiar with many of the current trials. Though it does appear that most gene therapies are starting with easier and more accessible targets with a higher risk/reward ratio than a prophylactic against cancer.

Also, though it may seem obvious that having more tumor suppressor proteins like retinoblastoma protein [1], or Brca [2] or p53 [3] should be a good thing, I don't think it is always so straightforward. On the other hand, elephants do have 20 copies of p53 - and they have way more cells than humans without getting a proportional amount of cancer.

Also if you think about 'where' you'd have to target the excess protein with the therapy it would be 'everywhere that might develop cancer', which is a much more difficult target than asking where to put the gene that encodes for hemoglobin.

tldr; we'll get there, but we're starting with more mechanistically determined/constrained problems before we move on to a cancer prophylactic.

[1] https://en.wikipedia.org/wiki/Retinoblastoma_protein

[2] https://serotiny.bio/notes/proteins/brca1/

[3] https://serotiny.bio/notes/proteins/p53/

This is a really amazing visualization. One of the things that I like about it is that the visualization shows representative random, nonproductive interactions (the other molecules that bounce off of the enzymes, for example) and the enzymes don't typically "magically go straight to" where they're supposed to do, they're often doing a little bit of fumbling around first.

This is in contrast to a videos like this one, where the molecules move with a purpose: https://www.youtube.com/watch?v=bbbbbcAeCa4

or this one, where you watch protein helixes magically self-assemble by zooming into exactly where they needed to be - a process so statistically unlikely it boggles the mind,


Lovely. A similar style of animation to that of Drew Berry (also the sound-design is similar). I wonder if the rotations of the atomic chains are accurate: Drew Berry's visualizations show some of the translational thermal noise, but little rotational noise.

What's hard in these animations is to depict the mechanisms without using the motion-graphics equivalent of intentional stance ("x evolved for y"). Molecules don't have intentions, only electric fields (which are never shown) by which they move. That's not including quantum effects...anyway many molecular animations give the false impression that the molecules have agency.

Last week there was a study on ice baths. You can't really find clear signal for more than 24 hours. Info is moving to fast.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact