Science Untangled: “Do Unhealthy Placentas Dream of Electric Sheep?”

How we use gene expression to understand disease

​This issue of our Science Untangled will focus on how we can use molecular biology to understand more about disease. In iPLACENTA we are interested in the causes of pregnancy pathologies; the juice of the matter is really to understand what's different between a healthy placenta and an unhealthy placenta that will cause the mother to develop a disease, putting both mother and baby in danger. 

The title of this article was based on the great novel by Philip K. Dick, that inspired the movie “Blade Runner”.   Like the androids from Dick's novel that at a first glance are very similar to real human beings, similarly diseased placentas are often not too different than healthy ones. 

Or rather, the differences become obvious too late, once the disease has already taken hold and the patients are in danger. In Dick's dystopian fantasy, the police has devised an interrogation method that will put the androids in a corner and force them to reveal themselves. In molecular biology, when we want to better understand what is going on inside an organ or a tissue, we often interrogate gene expression, looking for clues.

​Gene expression
​Like any other tissue, the placenta is made up by different types of cells, each type with its own function that will contribute to the placenta working smoothly. All cells in our body contain the same, identical genetic material in the form of DNA, packaged in chromosomes inside the nucleus. 

​The nucleus works as a library, storing on shelves – the chromosomes – hundreds of different books, which are the genes. It is the DNA molecule that contains the actual sequence of the genes.

Each cell is different because it uses a different set of genes which instruct on how to perform different functions, just like a biology student and a history student read different sets of books.

When a gene is read, it is activated: it is first transcribed into a copy, the messenger-RNA, so as to ensure that the original is kept safely away on the shelf, such a gene is now expressed. Depending on the cell we are looking at, for each expressed gene we have a certain number of messenger-RNA copies, this number is what we refer to as the level of expression of the gene. Therefore, when we measure gene expression, we are indeed measuring the number of copies of each of these genes.

When there is a disease, the level of expression of genes and even which genes are expressed can change. By comparing healthy persons and patients with a disease we can start to spot some differences.  In a normal placenta for example the HTRA4 gene is expressed relatively at a high level. However, when the mother develops the pregnancy condition called preeclampsia, these levels go up by 50 times!

How do we actually measure gene expression?
In the human genome, there are more than 40,000 genes that have been identified to date (Willyard, 2018). To gain as much information as possible we wish to measure gene expression at a global level, for this purpose the two techniques generally used are microarray and RNA-sequencing. Below a quick overview to get a taste of both!
If you are interested in knowing a bit more about the nitty gritty of these techniques, then read on the next paragraphs, otherwise you can directly skip to the next section The outcome.

The aim is to measure gene expression levels, for as many genes as possible. First, we collect the RNA molecules from the organ/tissues that we are interested in studying and from the patients we want to analyse. These RNA-molecules correspond to the copies of the genes and generally tell us how much a gene is used. Before carrying on, we need to convert these RNA-mixes to a more stable kind of molecule, the DNA, in a process called reverse transcription. This step is very easy to carry out since many kits are available on the market and the scientists are just left to mix all the ingredients together (without screwing up!) and wait for this reaction to occur. These converted molecules are called copy-DNAs (cDNAs) and are equivalent to the original RNA-molecules we had.  Now we can either do microarray or RNA sequencing (or both, if we have the money!). 

​Microarray.

1. The first step is to label our cDNAs molecules attaching a fluorophore, a molecule that when is exposed to a laser will produce fluorescent light that we can measure.
2. Now we cover the microarray chip with the cDNA-fluorophore mixes. The microarray is nothing other than a small support with different quadrants for different samples, in the example we have only 2 quadrants and 2 samples. In each of these quadrants we have many slots that can be filled with the mix. Each of these slots has high affinity for a specific gene – like a magnet, slot1 will attract the cDNAs that are the copies of gene1 and will be filled with them; in each quadrant we have slots for as many genes as the microarray can analyse.
3. At this stage, all cDNAs from sample 1 and sample 2 will have found their corresponding slot.
4. Depending on the gene, we have different numbers of cDNA copies in each slot which means that we also have a different number of fluorophores. Shining the laser on the microarray causes the fluorophores to emit fluorescent light that we can see (here, in purple) and measure.
5. We can immediately appreciate different shades of purple, that correspond to different intensities of fluorescence; stronger purple spots correspond to slots that contain many cDNA-fluorophores, and therefore to genes that are expressed at high levels.

​
RNA-sequencing.

1. First, we analyse our cDNA-mixes with a sequencer, that will give us the actual sequence of each molecule. These sequences are called reads.
2. The next step is computing intensive and consists in collecting all the reads and working out from which gene they are coming from. In order to do so, we use the sequence of the whole human genome as a reference and we look for matches between the sequence of the reads and the genome in a process called mapping.  
3. Now we know from which gene each read is coming from.
4. Counting the number of reads that have been mapped to each gene we can estimate its level of expression.  We can easily see in the example, how gene2 that was expressed at high levels produced the highest number of reads.
5. Finally, we look at differences between samples and write up a list of genes that have different number of reads, gene2 in the example. 

The outcome
The final result of both techniques is a list of genes that are expressed at different levels in disease compared to healthy organs. Among these genes, some are found to be expressed at a lower level – down-regulated, while others are increased (like the case of HTRA4) – up-regulated.
 
What do we do with this info?
Our overriding goal is to understand what is happening in these organs and what is happening differently in disease. Now we have quite a detailed overview of all genes that are behaving differently, but how can we interpret this information?
Usually, from a gene expression study such as the one that we described here we get quite big lists of changed genes, from hundreds to thousands, and in a first instance looking at this list voice by voice would drive us mental pretty quickly. So, we look for clues (again!): we group and categorise these genes by the functions they are known to be involved in, performing a gene ontology analysis.  Thankfully, many user-friendly softwares and websites are available for this (I am a not-so-skilled-with-computer-stuff biologist, very thankful to the programmers on this one). 

The results of this analysis look something like the real-life example on the left, which comes from the work of Han et al. (2019). Now, rather of than a list of thousands of genes, we have a finite list of functions that appear to be affected by the disease, things like cell growth, metabolism, response to stress.
How does disease affect these functions? Or could it be that the changes we see in these functions are causing the disease? Time to put the thinking cap on and come up with new, original, (smart?) hypotheses to test!

Some final considerations
We have gone through ways we can use to identify and try to understand diseased organs (kudos to you for sticking it to the end!). Differently from the grim fate of those androids, who once identified would be certainly eliminated, our hopes for the unhealthy placentas are much brighter, we want to understand so to be empowered to develop novel new successful treatments.
“We need not to be let alone. We need to be really bothered once in a while. How long is it since you were really bothered? About something important, about something real?”
― Ray Bradbury, Fahrenheit 451.
 
References
Han, K. et al. (2019) ‘Genome-Wide Identification of Histone Modifications Involved in Placental Development in Pigs’, 10(March), pp. 1–11. doi: 10.3389/fgene.2019.00277.
Willyard, C. (2018) ‘Expanded human gene tally reignites debate’, Nature, 558(7710), pp. 354–355. doi: 10.1038/d41586-018-05462-w.

---

Author

Clara Apicella is an Early Stage Researcher of iPlacenta. Read her earlier blog post here.