Gene Editing for Dummies

Rewriting the human genome to cure disease and extend our lifespan

The 21st century is the era of genetically-modified humans. Advancements in customizable restriction endonucleases, the proteins responsible for editing the physical DNA molecules in our cells, culminated in the discovery of CRISPR in 2012. It’s difficult to overstate the impact CRISPR has had on gene editing research; it was Science’s breakthrough of the year in 2015 and landed its discoverers Jennifer Doudna and Emmanuelle Charpentier the Nobel Prize of Chemistry in 2020.

And there is a tremendous need. Despite their name, rare diseases have such breadth (over 7,000) that as many as 300 million people worldwide live with them. These diseases are primarily genetic. Other diseases such as cancer, which are often not hereditary, are still genetic in nature as mutations in cellular DNA enable cancer cells to divide indefinitely. All diseases (rare or common) have some genetic origin.

However, the hype surrounding CRISPR has not necessarily translated to widespread medical successes, at least so far. The first CRISPR-based gene editing therapy was recently approved for sickle-cell disease and beta thalassemia, but with a price tag of $2.2 million it’s unclear how widely available this treatment will be. It serves as a proof-of-concept, and researchers around the world are chasing more accessible CRISPR therapies to reach more patients.

If the cost does come down and the therapy proves safe in the long run, gene therapies utilizing CRISPR will expand to larger disease categories. There may even be an economic incentive to introduce preventative gene editing: prevention is almost always easier and more effective than treatment.

If you could decrease the risk you ever get cancer in your life by 50% with a preventative gene therapy, would you take it? In order to understand just how feasible this is, we must first learn how gene editing works in medicine today.

The National Institutes of Health initiated the Human Genome Project in 1990 with a simple goal: piece together the sequence of all 3 billion nucleotides (A, C, G, and T) that encode our lives within 15 years. The vision was (and still is) to uncover countless genetic predispositions to diseases from cancer to Alzheimer’s. The project concluded in 2003 with a (mostly) complete human genome, funded largely by the US government at a $2.7 billion price tag, under budget and ahead of schedule (!!!). It was perhaps the biggest scientific success story the US government had undertaken since the Apollo program.

The human genome project gave geneticists a reference map to work off of. It also developed many technologies necessary to sequence large pieces of DNA. Several companies realized the potential that genetic sequencing could have in medical practices and biotechnology if it was faster and more cost effective. They built on the project’s work and developed novel techniques that slashed the cost and the time it takes to read DNA base pairs. Today, a human genome can be sequenced for around $1,000 in material costs in a matter of hours. From $2.7 billion to $1,000 in 20 years.

The astonishing progress in slashing the cost of sequencing a human genome.

In parallel with the development of genomic sequencing, scientists were also investigating proteins that “cut” DNA in sequence-specific sites. Research labs created artificial restriction enzymes that can be tailor made to target any particular DNA sequence. These customizable DNA cleaving proteins opened the door to limitless gene engineering and spawned numerous clinical trials to treat diseases from HIV to cancer.

The challenge here, though, is a new protein needs to be synthesized for each target. This long-winded process takes months to design, construct, and test each individual designer nuclease.

Enter CRISPR. This innovative technology popularized by bacteria uses a single enzyme, the Cas9 nuclease, to cleave viral DNA at a specific recognition sequence. In order to find that recognition sequence, the Cas9 enzyme is chaperoned by a short guide RNA. The guide RNA binds to Cas9 and takes it to a complementary DNA sequence present in the genome.

(Quick aside: CRISPR is the technique used by bacteria to recognize viral DNA and respond to it. Cas9 (literally “CRISPR associated protein 9”) is the protein that physically cuts the DNA. I use both throughout this story, but keep in mind they refer to the same technology.)

The protein (Cas9) is the same, and the guide is customizable to a particular target sequence. It is relatively easy to make a single protein in large quantities, and relatively easy to synthesize short custom RNA molecules, overcoming the challenges of previous gene editing proteins.

So simultaneously, biologists discovered efficient and customizable DNA-breaking enzymes and geneticists provided a map of the human genome. This merger of genetic engineering ingredients inevitably led many prominent publications to declare this the beginning of the gene editing revolution.

How do you edit a gene?

To edit a gene, first you have to pick a gene to edit.

A genome database is the perfect resource for this. Databases such as Ensembl allow researchers to peruse the genetic code of dozens of organisms using genome wide sequence data collected over the last 20 years.

Let’s look at what the first FDA-approved CRISPR treatment targets: the BCL11a gene. Patients with sickle-cell disease and beta thalassemia have mutations in their beta hemoglobin gene (HBB) that cause their red blood cells to “sickle” rather than develop into the proper torus-like shape. Medical doctors and scientists have found that sickle cell patients are often symptom-free for the first year or so of their life while the BCL11a gene is dormant. The gene is suppressed in early childhood but turns on as time passes, shutting down anti-sickling fetal hemoglobin production. Thus, disrupting BCL11a can enable blood cells to continue to produce fetal hemoglobin beyond childhood and improve the symptoms of sickle-cell disease and beta thalassemia patients.¹

Our task is then to scan the entire human genome for the perfect guide RNA so Cas9 can only cut the DNA encoding BCL11a while leaving the rest untouched.

Just kidding. Let’s use machine learning to do that for us.

Since the discovery of CRISPR and the understanding that it will be a dominant player in gene editing for decades to come, teams around the world have built tools to pick the best guide sequence for any given gene in many organisms. One of the best is CRISPRon, developed at the University of Copenhagen and led by Professor Jan Gorodkin. The algorithm uses machine learning to scan the human genome and find the optimal guide sequences that are specific to an inputted target gene, leaving off-target DNA untouched.

We can use these tools to design a guide RNA specific to the BCL11a gene, or any other gene of interest, to lead our Cas9 protein. For any highly motivated readers, you can get a CRISPR guide RNA synthesized by a company like IDT for $100 or so and begin your experiments. We are ready for CRISPR’s Cas9 to edit the BCL11a gene.

Cas9 follows the guide sequence to the BCL11a gene and breaks the DNA…

…what happens next?

Cells hate when their DNA is broken. Our cells are constantly bombarded by environmental factors that damage our DNA, like ultraviolet light from the sun, and have evolved numerous repair mechanisms to protect us from malfunctioning genes.

Immediately after the DNA breakage in the BCL11a gene, proteins are shuttled to the site to evaluate the damage. One thing they are looking for is a repair template. If there is a DNA sequence that shares a lot of similarity to the damaged sequence, the cell can use it to repair the damaged gene to its prior form. For instance, if one chromosome has damage in a gene but the other is untouched, the cell can use its homologous chromosome to repair the altered one. This process is known as homologous recombination: a homologous (i.e., complementary) sequence is used to recombine the DNA sequence.

With gene editing using CRISPR, chances are that both chromosomes will be edited. The cell therefore does not have a natural repair template to use. It still needs to fix the break, though.

Like answering an exam question without studying, the cell makes it up as it goes. Enzymes chew back the damaged DNA bases, removing them. Another protein called DNA ligase 4 will  “repair” the DNA, fixing the break that Cas9 caused. This process does not depend on any homologous sequences to replenish the damaged bases; those bases are lost in the process. This repair mechanism is called non-homologous end joining (NHEJ) and is known to be error prone due to the lost information. But, it works perfectly for situations like these. It fixes the DNA break and allows the cell to continue doing as cells do.

A render of DNA ligase 4 repairing DNA damage.

When delivering Cas9 with its guide sequence to a cell, the gene it targets will most often be edited and repaired with NHEJ. The loss of information will cause the problematic BCL11a gene to become “nonsense” most of the time, meaning it does not encode a functional protein. The gene will be “knocked out”.

So we use Cas9 to insert breaks in the BCL11a gene, and the cells attempt to repair the damage in an error-prone manner that disrupts its sequence. This process turns off BCL11a expression and turns on healthy fetal hemoglobin production, treating patients with sickle cell disease and beta thalassemia.

This, my friends, is gene editing. We have our formula to cure sickle cell disease.

Yet we haven’t covered a major problem: which cells are we editing? All of them???

No way. Adults have around 37 trillion cells. Targeting all of them is impossible with current technologies; the only technical way to do this today is early in embryonic development, when we are a single cell or small clump of dozens of cells in a test tube and before many genetic diseases like sickle cell present themselves.

For sickle-cell and other blood disorders, we can take advantage of blood stem cells present in our bone marrow. All of our 500 billion blood cells produced every day come from just 10,000 stem cells. So to make healthy blood in patients with sickle-cell, we just need to edit their blood stem cells to disrupt the BCL11a gene. It is still a difficult process, to be fair, but much simpler than modifying all 37 trillion cells.

The process looks something like this: The patient has a bone marrow sample taken to collect as many blood stem cells (known as HSCs, or hematopoietic stem cells) as possible. These HSCs are grown in a laboratory setting under intense sterile conditions. Once a sufficient number of these cells are ready, scientists will briefly electrocute the cells to form pores in the cell membrane that the Cas9 protein and its guide RNA can traverse through and edit the BCL11a gene. The cells are allowed to recover, then the scientists will probe the cells to validate that Cas9 was able to disrupt the BCL11a gene in enough cells, often through genetic sequencing (thank you Human Genome Project yet again). The edited HSCs will be grown to reinfuse back into the patient. 

While this is ongoing, the patient is undergoing intensive chemotherapy to kill off their residual blood cells and HSCs. The edited HSCs will then be transplanted back into the patient’s bone marrow. The patient will be monitored for any signs of immune rejection to ensure the cells are engrafting and beginning to produce healthy blood cells. This whole process from start to finish can take 6 months or more.

Given all these challenges, it’s a marvel that scientists at Vertex Pharmaceuticals and CRISPR Therapeutics were able to release a potentially curative therapy using CRISPR technology within 12 years of its discovery.²

The celebrations over CRISPR and its first FDA-approved therapy have all been tempered by its price tag. At $2.2 million, it's one of the most expensive therapies available worldwide and, as mentioned earlier, leaps and bounds from entering first-line treatment options for more common diseases.

As alluded to earlier, CRISPR therapies are theoretically more cost-effective than previous gene therapies. A vial of Cas9 nuclease for laboratory testing can be purchased for $80, while the corresponding guide RNA sequence can be synthesized for around $100. The media, aka the food needed to grow HSCs, is available for $451. The materials to electroporate the cells can be bought in kits for $30 per experiment. Laboratory equipment such as cell culture incubators, biosafety cabinets, the electroporation device itself, refrigeration and liquid nitrogen for long term storage are all notoriously expensive, but these are relatively fixed costs as they can last for many years.

We can spec out the materials to be <$1,000 for a therapy that costs 2,200-times that. So, what’s the catch? Yet another instance of big pharma profit-motivated greed?

Well, maybe. But the above costs are not telling the full story.

For starters, these prices are for research-grade materials that are explicitly not allowed for any therapy that will be injected into a person. They are also the smallest size available, great for laboratory experiments but not enough to edit and grow a clinical scale number of cells. A corresponding vial of clinically-compliant Cas9 protein at a clinically relevant size costs more than $10,000.

A clinical therapy also cannot use any ordinary laboratory. It must be processed in a certified clean room by technicians and quality control managers with years of experience. These employees need to devise standard operating procedures for these processes, which in the case of a novel CRISPR therapy need to be brand new and still meet the FDA’s rigorous criteria. They require payroll, of course, and having such clean room experience commands a high price.

So the $1,000 materials cost of research-grade experimental materials is a bit of a façade. It represents more the technical lower bound of material costs associated with the therapeutic development, removing the fixed instrument costs and the labor costs. It shows the potential for CRISPR therapies to be widely available, but also the long road we must take to get there.

What does the future have in store?

Sickle-cell disease is the poster child disease for CRISPR-mediated gene editing. It's a condition with a well-established causative mutation, with accessible cells for editing, and with few alternative options for patients with the disease. It's no wonder this is the first disease to receive FDA approval for a precision CRISPR gene editing product.

However, the cost must come down for CRISPR to live up to its potential.

While rare diseases are, well, rare, there are so many different diseases that as many as 1 in 10 people in the US have a rare genetic condition. And most rare diseases have fewer addressable patients than sickle-cell, meaning the treatment pricing to recuperate R&D costs needs to be even greater.³

One clear avenue is to simplify the whole therapeutic manufacturing process; rather than collect patient cells, edit them in an off-site lab, and reinfuse them in a months-long process, what if patients could be injected with the editing cargo right in their hospital bed? Give the patients an infusion of CRISPR particles that find their way to the bone marrow. This is being actively investigated by many labs, including Jennifer Doudna’s, one of the discoverers of CRISPR.

Another solution could be to find diseases with larger patient pools that can be treated or cured with genetic engineering. Reaching economies of scale in manufacturing these therapies could bring the cost down substantially. Combining both - in-person gene editing with scalable manufacturing - could bring gene editing to mainstream medical care. We just need to pick a widespread, devastating disease treatable with a single genetic target.

Hmm.

What about cancer? Like, all cancer?

Cancer was previously thought of as a large mammal disease. The larger an animal and the longer it lives, the more cells it has that can mutate into cancer and the more time they have to become cancerous. This logic follows quite well. In humans, body weight and lifespan are both correlated with increased cancer incidence. Yet African elephants, which weigh up to 10,000 pounds and live up to 60 years, do not get cancer.

African elephants.

What gives?

Consider the p53 gene. Also known as “the guardian of the genome”, p53 is one of the proteins responding to DNA damage and repair in our cells. If DNA repair is not possible, p53 can stop the cell from dividing at all, or even kill the cell to prevent cancer formation. More than 50% of cancers have mutations in p53; it is so damn good at its job, cancers pretty much have to shut it down in order to develop.

Humans have one copy of the p53 gene in our genome. African elephants, in contrast, have 20 copies. If p53 is mutated in humans, we’re screwed. But in elephants, they have 19 other backups to rely on. With p53 mutations so heavily implicated in cancer cases, adding a few extra copies of this gene could theoretically prevent half of cancer incidences.⁴

Given a theoretical patient pool of more than 300 million in the US and 8 billion worldwide (also known as the entire human population) for preventative cancer therapy, the manufacturing and labor costs could shrink much closer to the theoretical minimum as processes for manufacturing and administering CRISPR associated therapies are optimized.

I bet health insurance companies would pay for it, too. The lifetime healthcare costs for cancer patients is over $200,000. With cancer incidence rates approaching 50% in the US, if there was a preventative therapy that halved its occurrence and cost less than $100,000 it would 1) save health insurance companies a tremendous amount of money and, more importantly, 2) add countless years of healthy lifespan to the human population.

So, back to the question posed at the beginning of this story: would you cut your chances of ever getting cancer by 50% with a preventative gene therapy?

This may seem like a wild question at first, but perhaps after reading you realize this is a real possibility. CRISPR therapies exist today, and their applications are broadening every day. And I believe we should start talking about it now so we are prepared for a world where it is attainable. 

There are concerns regarding gene editing en masse with our current knowledge, though. Off-target editing has the most damaging potential impact on the patient. Previous and current gene editing techniques using engineered viruses are known to incorporate randomly in the genome, which can cause cancer if they mutate “oncogenes” or cancer-causing genes. CRISPR promises to be more precise but caution must be taken to minimize the potential for off-target editing. And we still don’t know the long term safety profile of CRISPR-edited cells in patients.

With these limitations considered and the future uncertain, the progress made so far in precision gene editing is nonetheless remarkable. As a novel technology, CRISPR/Cas9 went from discovery to clinical therapy in around a decade. The therapy may be a lifelong cure for patients with sickle-cell disease, removing chronic pain and lengthening lifespans for thousands of people in the US and potentially millions worldwide. Another CRISPR therapy is in clinical trials to treat blindness in children with inherited vision loss.

Alongside the approval of Vertex’s sickle-cell therapy, many research groups are moving forward searching for better answers to combat the limitations of this initial groundbreaking product, and countless labs are working on gene editing cures for hundreds of genetic disorders plaguing millions of people worldwide. Including the original discoverers of CRISPR.

Gene therapy has the potential to treat countless genetic diseases this century. And our growing understanding of its long term safety and effectiveness will allow us to use it more frequently as a first-line therapy. The potential for gene editing therapies is enormous. There are hundreds of diseases curable by simple changes in our genetic code. We may use this technology to write our own chapter in the epic story of humanity. The directed evolution of Homo sapiens.

1. It is actually slightly more complicated than this. The BCL11a gene is rather important for white blood cell development, notably of antibody-producing B cells (BCL11a does stand for B-cell lymphoma/leukemia 11A, after all). So rather than completely disrupting BCL11a expression, the CRISPR therapy targets a red blood cell specific region of the gene. This disruption allows BCL11a to continue functioning as normal in other cell types (like B cells), but is functionally inert in red blood cells, which are the main carriers of hemoglobin and thus the disease-relevant cells in sickle-cell disease.

2. Let’s put this 12-year clinical timeframe into perspective. A similar class of engineered cell therapy, CAR T cells, was first described in research in the 1980’s and generated in 1993. The corresponding therapy was approved by the FDA in 2017, 24 years after its publication. Induced pluripotent stem cells, a novel class of stem cell, were discovered in 2006 (6 years before CRISPR) and similarly yielded its discoverers a Nobel prize. Yet the first clinical trials with these cells have only begun in the last year, and an FDA approved product is not anticipated until the late 2020’s at the earliest. Compared to similar classes of novel therapeutic development, CRISPR has at least halved the time from discovery to clinical approval. That is lightning speed.

3. The cost of a therapy generally boils down to two things. First is the manufacturing of the treatment. Between different genes, we are still using CRISPR and its Cas9 protein, so the manufacturing costs should be roughly the same, save for some manufacturing optimization that can decrease costs. The second thing is the research and development costs to bring a therapy to market. It is generally accepted that it requires at least a billion dollars to bring a treatment from idea to FDA-approved product. This cost needs to be recuperated by selling the product for more than it costs to manufacture it. For a therapy targeting a common condition like arthritis, that cost is spread thin across tens of millions of prospective patients, so the pharmaceutical company only needs to make like $100 per patient. For rare genetic diseases, by definition that cost is borne by a smaller patient pool. So with something like 10,000 prospective sickle cell disease patients in the US eligible for CRISPR therapy, the company needs to make closer to $100,000 per patient. For even rarer diseases, that can push the cost to more than $1 million per patient on top of the manufacturing costs. These are all approximations, but it shows that for rare diseases, the pricing needed to pay for the R&D cost is significantly higher.

4. We described the technology and biology behind “knocking out” a gene in our genome, or removing its protein expression. We have not discussed how to “knock in” a new gene (or genes) into our genome. However, we can use the processes we described for DNA repair to guide us there. It starts by delivering the same Cas9 protein and guide RNA, in addition to a repair template with our “knock in” gene. The Cas9 protein induces a DNA break, and the cell searches for a repair template. We conveniently provide a repair template that allows the cell to repair the breakage, while also including some extra DNA encoding the gene we want to “knock in”. Generally speaking, “knocking in” a gene is less efficient than “knocking out” one (not all DNA breaks using Cas9 will be repaired with our provided repair template). Hence why the CRISPR therapy “knocks out” BCL11a rather than “knocking in” a healthy copy of the beta hemoglobin gene. But this technique of “knocking in” genes has been utilized in laboratory research extensively to engineer cells and study their behavior, and is entering clinical trials to validate its medical use.