In a recent paper by Graham and Root[1] at
MIT/Broad the authors provide an interesting list of CRISPR facilitators. It is
akin in many ways to the old Sear catalog. For folks who may be much too young
to remember the Sear catalog, if one lived
quite a distance from a large store, or if you just wanted to spend time
looking at what you could get, Sear sent out 3” thick tissue paper catalogs.
They had page after page of “stuff”, and Amazon before Amazon.
In the paper by Graham and Root one finds a much thinner
version of this for CRISPRs, but with activating the URLs attached one can get
the equivalent feeling. A complete set of Sears Craftsman tools, but for
assembling genes via a piece by piece method.
Recall that CRISPR is that targeting sequence of an RNA to
attach to a specific part of a gene so that the Cas9 protein can cut the gene
at the end of the PAM sequence. Also recall that we can use different proteins
from Cas9 to make offset cuts. Cas9 makes cuts opposite one another which often
leads to possible reconnection problems.
The CRISPR-Cas9 combination is but one of many that can
apparently be used. Changing Cas9 for other proteins lead to altered cutting
mechanisms which we have discussed previously. There are in the Graham and Root
paper four fundamental applications. They are:
1. Cutting or Knock Out (KO) where a gene sequence is cut
from a chromosome.
2. Pasting or Knock In (KI) where a gene sequence is added.
3. Inhibition where a gene expression is inhibited by
creating a block to a transcription factor.
4. Activation where an activator is effected by the CRISPR
Cas9 action.
In some sense types 3 and 4 are also done by methylation or
indirectly via miRNAs.
Graham and Root first discuss design factors and they state:
The following considerations and guidance apply to many
types of CRISPR-based experiments:
The delivery method for Cas9 and sgRNAs can be
transfection or transduction. Viral transduction is required for pooled
screens. More consistent activity can be provided by selection of
Cas9-expressing, CRISPR-active cells. CRISPR activity can be variable across
cell types and should be experimentally confirmed case by case.
The delivery mechanism is a critical factor. How does one
get the combos into a cell? Some recent work indicated that a cell can make its
own Cas9 and then inject the CRISPR. The challenge is also targeting specific
cells. In some areas one has looked at this approach almost as a weaponized
system and the delivery mechanism is as critical as the actions themselves.
One should select sgRNAs to maximize the likelihood of
high activity and specificity. The current state of knowledge provides useful guidance
for selecting target sites and sgRNAs, but predictions of efficacy and
especially of specificity are currently far from perfect. Table 1 describes
tools now available to assist in this process. New tools and strategies are
arising frequently as the understanding of CRISPR technology improves.
The sgRNAs used to select the gene sequences are not always
that selective and have less than perfect specificity. This can be complicated
in a wild type environment where the genes may be present in a certain large
percentage but intra-species variations are possible.
Multiple sgRNAs per target gene (typically ranging from
three to eight) should be employed wherever possible, first, to provide more opportunities
to achieve the desired on-target modification and, second, to evaluate
concordance of the phenotypic effects of multiple independent reagents to
prioritize results most likely to be on-target — that is, causally linked to
the intended genetic perturbation.
Toolkits for better targeting and specificity are essential.
Validation of genetic models and phenotypes is essential.
Confirmation of on-target efficacy is important for selecting good cell clones
to use for subsequent experiments and to establish the specific gene edits
produced. Experimental assessment of off-target effects of sgRNAs can also
inform clone selection.
The targeting is still not perfect. It is also not clear
what effects such epigenetic factors such as methylation will have.
They do discuss the usefulness of this technique in
examining the functions of genes by KO analyses. However as we learn more about
genes we see that are a mass of interconnected strings and the one gene one
function model of Mendel has outlived its usefulness.
They conclude by stating:
One major goal is to achieve more efficient, predictable editing.
If it were possible to convert every cell in a population to the desired
genotype, the painstaking work of selecting and characterizing individual
clones would be reduced or eliminated. This would make it feasible to engineer large
numbers of clonal cell lines, or even to engineer specific alleles at a
screening scale. It would also make it far more efficient to produce cells with
multiple edits. One approach is to re-engineer Cas9 for desirable
characteristics, including altered PAM sequences, better packaging into virus,
better binding and cutting efficacy and higher specificity. The hunt is also
under way for better type II Cas9 proteins or other type II CRISPR proteins that
might possess performance advantages, or to provide altogether new activities.
The adoption of new CRISPR systems might necessitate new studies to determine their
on- and off-target behavior and ideal design parameters.
This paper is well worth using since it is a balance of what
is achieveable and provides and exceptionally useful tool box for those working
in the area. The only issue is that the area is changing so rapidly that the
tool box may have to be updated frequently.
[1] http://www.genomebiology.com/2015/16/1/260
Graham and Root, Resources for the design of CRISPR gene
editing experiments, Genome Biology (2015) 16:260