We are not mice

One of the reasons why the results of studies conducted on mice cannot be transferred to humans without looking back

One reason mouse studies often don’t translate to humans very well David Gorski, Science-Based Medicine

Translated by Alexander Gorlov, XX2 century
See the links in the translation or in the original of the article.

Mouse models are often used for preclinical modeling of human diseases, but drugs successfully tested on mice are recognized as suitable for humans in about one case out of ten. Why? A new study helps to answer this question, which compares the expression of genes in human and mouse brain cells.

Is there something in the scientific literature that makes many researchers, in particular doctors, lose their minds? Perhaps there is, and this is irrepressible praise as the locomotive of scientific progress of preclinical research conducted on cell cultures and animal models, and these studies are praised even when the method of treatment studied during their implementation has not yet been transformed into a method of treating people. Don't get me wrong; sometimes the study of mice leads to an important breakthrough in medicine. However, something else is much more common: treatments that are effective for mice cannot be transferred to humans, because they do not treat people at all or treat very poorly. My favorite example is angiogenesis inhibitors, which I studied at the beginning of my scientific career. Back in the late 90s of the twentieth century, angiostatin and other angiogenesis inhibitors - drugs that can prevent the formation of new blood vessels in a tumor – were advertised as a cure for cancer. The fact is that the development of cancer, at least those of its most dangerous types, depends on its ability to force the body to generate new blood vessels with a diameter of more than two millimeters for the continuous growth of the tumor in order to satisfy its insatiable need for oxygen and nutrients. In addition, impressive results were obtained on mice. My wonderful mentor, the late great Judah Folkman, successfully cured them of cancer. He could even cause them to "hibernate" the tumor – such a state of it when, having shrunk to a tiny size, it remains completely inactive for an indefinite time. And all this with the help of angiostatin and endostatin, two inhibitors of endogenous angiogenesis, which he isolated from the urine and blood of mice. Unfortunately, in the treatment of cancer in humans, angiogenesis inhibitors work, but very poorly, without giving a quick impressive effect, which is why they are used along with many other drugs in order to gradually improve the patient's condition. At that time, Dr. Folkman used to have a sarcastic joke: "If you have cancer and you are a mouse, we will certainly help you."

So what's the matter? Why are many drugs that successfully treat mice ineffective in treating humans? Take, for example, neuropsychiatry: here, attempts to predict the effectiveness of new methods of treating people based on data obtained with the help of mouse models turned out to be particularly deplorable. To get an answer to the question we are interested in, let us turn to the results of a recently published study that showed that one of the reasons for this inefficiency, at least when creating drugs intended for the treatment of mental and neurological disorders, is the previously ignored difference in the functioning of the brain. Before analyzing this scientific work, let me give the floor to Sharon Begley, who wrote about it for STAT News:

"In the name of science, laboratory mice have to suffer a lot, but this is often compensated, albeit for a short time, by the discovery of effective, almost miraculous remedies for diseases that kill people. Unfortunately, experimental drugs against Alzheimer's disease, schizophrenia and glioblastoma, which cured millions of mice, did not cure a single person, and this sad fact suggests that when it is necessary to replace people when testing new treatments for many brain diseases, mice are very lousy models.

According to a report published on Wednesday, scientists were able to establish the main reason for the different effects of drugs on mice and humans: the human cortex contains cell types that are absent in the cortex of mice, and, most importantly, the activity of the key genes of these human cells differs significantly from the activity of the key genes of mouse cells.

As part of the most detailed classification of human brain cells to date, conducted by a team of scientists the size of a symphony orchestra, brain cells are sorted not by their shape and location, as has been customary for decades, but by what genes they use. Along the way, important discoveries have been made, and among them is this: mouse and human neurons, which were considered the same under standard classification forms, can significantly (ten or more times) differ in gene expression for such key components of the brain as neurotransmitter receptors." 

I do not know if the team of scientists was really "the size of a symphony orchestra" (according to my calculations, it included about 63 specialists plus or minus two), but this kind of research really requires a large research team composed of representatives of several universities, such as the University of California at Davis (University of California, Davis), University of Washington (University of Washington), Columbia University (Columbia University), University of California in San Diego (University of California, San Diego), as well as a number of institutes. However, as can be understood from the above quote, the main thing is not this, but the fact that in mice and humans, neurons and circuits connecting different areas of the brain, even having striking anatomical similarities, can be very different at the level of gene expression.

And how did the classifier researchers work? First of all, they collected biological material: firstly, control material - the brains of deceased male and female donors aged 18-68 years who did not have neuropsychiatric or neurological diseases, and secondly – brain tissues of patients who underwent surgery for epilepsy. In all this material, the greatest scientific interest was the middle temporal gyrus (MTG), a part of the cortex that often undergoes resection in the treatment of epilepsy, and therefore allows you to compare the control material with the material taken from epileptics. The researchers then sequenced the single nuclei of the collected samples:

"Molecular classification of cell types has become possible thanks to single cell transcriptomics, which provides a metric for comparative analysis and stimulates efforts aimed at identifying the complete cellular composition of the mouse brain and even the entire human body. Sequencing of single cell RNA (scRNA-seq) of the mouse cortex demonstrates reliable transcription signatures of cell types and indicates the presence of about 100 cell types in the cerebral cortex. The complexity of extracting living cells from the human brain makes it difficult to apply scRNA-seq to this type of tissue. Another thing is the sequencing of single nuclei RNA (snRNA-seq), which allows for transcriptional profiling of cell nuclei contained in frozen human brain samples. It should be noted that the information available in the nuclei on gene expression is quite sufficient to distinguish closely related cell types with approximately the same high resolution as scRNA-seq, however, the depth of coverage used during early attempts to study the human cortex using snRNA-seq was not enough for the resolution to be similar to that given by scRNA-seq. what is obtained when working with mice. And now we have created reliable methods for classifying human brain cell types using snRNA-seq and compared human and mouse cortical cell types to identify conservative (similar) and divergent (divergent) properties."

To minimize the effort, it would be best to sequence the genome of single cells, but apparently extracting individual cells from the brain without killing them is a big technical problem; therefore, the authors used another excellent method: sequencing single nuclei. Its disadvantage is that the nucleus does not contain all the created matrix RNA (mRNA), because for translation (protein synthesis), mRNA is transported outside the nucleus. However, in this case, as earlier studies have shown, sequencing of single nuclei makes it possible to detect differences in gene expression between different types of brain cells no worse than sequencing of single cells. In general, RNA-Seq is a new generation sequencing method that has made available the sequencing of each mRNA sequence from those that can be found in any RNA isolate from cells or tissue; currently, the method has been developed in such a way that it allows sequencing on RNA isolates of single cells and nuclei. 

Now let's consider the scale of the research conducted by classifiers. The nuclei were isolated from eight donor brains, with the majority coming from postmortem donors (n = 15,206), and the smaller (n = 722) from epileptics (fragments of the fifth MTG layer removed during neurosurgical procedures):

"In total, 15,928 nuclei passed quality control: 10,708 from excitatory neurons, 4,297 from inhibitory neurons and 923 from non–neuronal cells. Nuclei from each broad class of cells were iteratively clustered, as described below (see "Methods"). Clusters, as a rule, remained stable when using various iterative clustering methods, their difference from their nearest neighbors amounted to at least thirty differentially expressed genes and at least one (and often more) binary marker." 

Translation: After sequencing, the researchers noted that different cell types can be reliably identified using various mathematical clustering methods used to analyze genomics data.

The purpose of the study was to develop and test methods and a system for using snRNA-Seq to catalog MTG cell types, and later other parts of the brain. Therefore, if you are not a neurologist or a neuroscientist, there is no need to disassemble all the documented features of cell types and their locations. However, a number of observations made should be noted. For example, this: excitatory neurons turned out to be distributed much more widely than expected, while most types of these neurons are not localized within a single layer. In layers L2 and L3 there are mainly three types, whereas in layers L3 – L6 there are as many as ten. Such heterogeneity, the authors of the study noted, "leads to the conclusion that to determine the type of neuron, it is not enough to establish its anatomical laminar location, but whether this is typical for the entire human cortex or only for MTG is still unclear."

When the researchers compared these gene expression profiles of human single nuclei with similar profiles of mouse single cells in the same brain region, they noted several differences. It was found that the neurons of mice and humans, which were considered to be the same based on anatomical, structural and histological features, as well as in accordance with standard classification schemes, may actually have a tenfold or even greater difference in gene expression for the synthesis of very important proteins, such as neurotransmitter receptors. (Neurotransmitters are peptides used by neurons to communicate with each other).

For example:

"Comparison of homologous types showed a mixture of conservative and divergent expression. The type of Sst Chodl (in humans – Inh L3-L6 SST NPY) generally demonstrates conservative expression, but in 18% of genes the expression turned out to be highly divergent (a conservative approach to understanding divergence was used here: the difference should be more than tenfold), including many marker genes. The picture is similar in oligodendrocyte progenitor cells (OPCs): generally conservative expression and 14% of genes with high divergence of expression. Two thirds of the analyzed genes (9748) showed divergent expression in at least one of the 37 homologous types, and many had changes in expression within the same type or class. Cell types that are not neurons showed the most divergent expression (3643 genes with more than a tenfold difference), which speaks in favor of increased evolutionary divergence of non-neuronal expression patterns of human and mouse organisms."

Translation: when comparing mouse and human genes, almost 20% of the genes showed very different expression (more than a tenfold difference) and at least two thirds of the genes showed different. This difference was most noticeable in the gene encoding serotonin receptors:

"Serotonin receptors demonstrate highly divergent expression for the two species studied: four of the seven G-protein coupled receptors and both ionotropic receptor subunits (HTR3A and HTR3B) were among the 10% most divergent genes (Fig. 6e). The most divergent gene families include neurotransmitter receptors, ion channels, extracellular matrix elements, and cell adhesion molecules. Of the top 3% of the most divergent genes (additional Table 5), collagens COL24A1 and COL12A1 and glutamate receptor subunits GRIK1 and GRIN3A were expressed in different cell types of the two species, and it was confirmed that they have different laminar distributions in humans and mice (Fig. 6f, g). The cumulative effect generated by many differences in cellular gene patterns with well-characterized roles in the transmission of neural signals and ensuring neural connectivity, of course, it cannot but cause many differences in the functioning of the human cortical circuit."

Serotonin (5-hydroxytryptamine, or 5-HT), as it is known, is a molecule with many functions performed in different parts of the body. In particular, serotonin can function as a hormone, growth factor, or neurotransmitter. For example, it accumulates in platelets, which release it when binding to a blood clot, and when the concentration of serotonin is high, it causes narrowing of blood vessels. (It is interesting to note that at low concentrations it acts as a vasodilator). In addition, serotonin promotes wound healing. However, in the brain, it is primarily a neurotransmitter that affects, as scientists believe, mood, sexual activity, appetite, sleep, memory, learning ability, temperature regulation and some forms of social behavior. Serotonin has gained the greatest fame (at least among the general public) as one of the factors of depression – thanks to a whole class of drugs called selective serotonin reuptake inhibitors (SSRIs) and are used to treat social phobia, anxiety and panic disorders, obsessive-compulsive disorders (OCD), major depression, irritable bowel syndrome (IBS) and eating disorders.

As a rule, when two neurons communicate, one of them releases certain neurotransmitters, such as serotonin, into the space between them (synapse). A neurotransmitter diffuses from a given (presynaptic) neuron through a synapse to a postsynaptic neuron and binds to its receptor, thereby activating a signal in the postsynaptic neuron. After the signal is sent, neurons, using transporter proteins, get rid of the presence of an additional neurotransmitter in the synapse. As soon as the neurotransmitter bound to the receptor is released, the transporter proteins transfer it back into the cell. SSRIs inhibit this activity of specific serotonin transporters, which allows the latter, lingering in the synapse, to stimulate the postsynaptic neuron receptor for longer.

In connection with this circumstance , the authors of the study , during the discussion of the results obtained , note:

"Our results demonstrate the divergence of gene expression of the studied species in homologous cell types, which is shown both at the level of single genes and at the level of the structure as a whole. These differences are probably functionally significant, since divergent genes are involved in the binding of neurons and in the transmission of signals, and since many markers of cell types exhibit divergent expression. It is noteworthy that serotonin receptors occupy the second place among the most divergent gene families, which makes it difficult to use mouse models to study many neuropsychiatric disorders associated with serotonin signaling."

And in conclusion:

"These observations set a quantitative framework for the discussion of whether the human cortex differs from the cortex of other mammals, revealing the basic transcriptomic similarity of cell types, disrupted by differences in proportions and gene expression in different species that can affect the functioning of microcircuits. In addition, the results obtained help to resolve the paradox of the unsuccessful use of mice for preclinical studies, despite the presence of a conservative gene structure in mammals, and emphasize the need to supplement the analysis of the brain of model organisms with the analysis of the human brain. The scale of the differences between humans and mice is such that in order to study many aspects of the structure and functions of the human brain, a similar profiling of the brains of non-human primates much closer to humans should be carried out. In addition, the improved resolution made possible by molecular technologies offers great prospects for accelerating the mechanistic understanding of the evolution of the brain and its diseases."

And here is the opinion of lead researcher Ed Lein from the Allen Institute for Brain Science in Seattle, published in STAT News:

"All the drugs created affect receptors or other molecules," said neuroscientist Ed Lane of the Allen Institute for Brain Research in Seattle, who led the study, the results of which are published in the journal Nature. “If in human cells, unlike similar mouse cells, the neurotransmitter receptor that you hope to target is not used, then your drug will fall into the wrong chain." Therefore, it will not be able to have the same effect on humans as it did on laboratory rodents."

So, although the mouse and human brains have basic similarities, this study unexpectedly showed that in the model, which is a mouse version of the structures of the human brain, neurons use different neurotransmitter receptors significantly differently than in the original. The results of this study will help scientists better interpret the results of studies conducted on mice, and, hopefully, apply them more accurately in relation to human neurophysiology. Meanwhile, after learning about some research, check if it was conducted only on mice. There is even a @justsaysinmice feed on Twitter, the main purpose of which is to draw attention to widely advertised studies of this kind.

And while we are waiting for animal model studies to become more predictable in terms of the applicability of their results to humans, it makes sense to follow @justsaysinmice.

Portal "Eternal youth" http://vechnayamolodost.ru