What are the types of cell functions that you seek to interrupt with microRNAs? What tools and techniques do you use to elucidate the targets and roles of particular miRNAs?
We are using microRNAs to replace the function of a tumor suppressor.
We perform assays that analyze how miRNAs modulate cancer-related processes in vitro, such as cell proliferation, cell cycle, soft agar assays, as well as invasion, and migration assays using miRNA mimics and antagomirs from Ambion. We have also carried out tumor formation studies in a mouse lung cancer model, employing adenoviral vectors that overexpress miRNA in vivo.
Furthermore, we have been successful using the C. elegans animal model to determine how miRNAs function together with their targets to regulate cancer progression pathways related to cellular growth and differentiation during development. We are using RNAi suppression screens in miRNA mutant backgrounds, miRNA sensor constructs to determine areas of miRNA activity versus patterns of expression, in vivo miRNA sponges, transgenic miRNA overexpression constructs, and miRNA promoter::gfp fusions to study temporal and spatial miR expression patterns in vivo.
We are studying miRNA targets that may play roles in cell cycle, stress response, or transcription/translation regulation. We use double-stranded sRNAs and plasmid-based shRNAs for mimics, and LNA-DNA mixed oligos as inhibitors. For functional/mechanistic assays we use luciferase assays, real-time PCR, Western blot, and Northern blot assays. To enhance our studies, we also developed a highly sensitive Northern-blot protocol termed LED that uses LNA, EDC-cross linking, and DIG; LED is primarily used for detecting small RNAs that are very difficult to detect without manipulating the RNAs, as is done with other approaches that use amplification of the RNA.
So far we have employed miRNA to interrupt the growth and proliferation of tumor cells. We used shRNA, miRNA mimics, and inhibitors, as well as site-specific reporter vectors and expression profiling, to elucidate the direct interaction of miRNA with its targets.
One of the things we are seeking is to increase lymphoma cells’ sensitivity to chemotherapeutic drugs by modulation of miRNAs. For transient expression in suspension cells we use electroporation. The efficiency of transfection is really good and very satisfactory. In lymphocytes we see a 5- to 10-fold increase in miRNA levels by electroporation of miRNA mimics, which is what we would like to see in other cell types.
Infection of lymphocytes is challenging, and for stable expression we’ve used the miRZip anti-microRNA lentiviral system from System Biosciences for inhibition of specific miRNAs. The efficiency of delivery is very good, but we have some problems detecting the inhibition of the targeted miRNA.
We are currently studying the roles that mi-RNAs play in angiogenesis, proliferation, and cell death of cancer cells. Alteration of tumor microenvironment and enhancement of response to therapeutic drugs are among our main interests. We work with both solid and liquid tumors and a fair number of neoplastic phenotypes, and we look at a broad spectrum of expression patterns. We use microRNA mimics, and inhibitors from Dharmacon, and they work great.
But, when we put the same mimics in adherent cells we see a 10,000-fold overexpression, which is a concern for us and for journal reviewers who worry about artifacts at those supraphysiological levels of expression. Whenever possible, we use retroviral and lentiviral delivery systems for adherent cells.
We [Drs. Sotillo-Pineiro and Thomas-Tikhonenko] cannot really use qRT-PCR and claim that our microRNA of choice disappeared, because the way miRZips work is that they presumably sequester and stabilize inactive miRNA, rather than causing its degradation. Indeed, System Biosciences’ technical support will tell you to do functional assays to clarify the issue. But the reason we use miRZip is because we want to validate targets and their functions. So it’s a vicious cycle.
We need to have an established target to validate miRZip, but to know what a good target is you have to use an antisense inhibitor. It is a big problem for us, and we haven’t really solved it yet. However, canonical antisense inhibitors (Dharmacon) do reduce steady-state levels of microRNAs. We also use Alnylam’s antagomirs (a class of chemically engineered oligonucleotides used to silence endogenous microRNA) that easily enter the cell, as they are linked to a cholesterol molecule. With this strategy we find the microRNA levels to be greatly reduced, as evidenced by qRT-PCR.
Which microRNA profiling techniques do you use to study tumor cells and why?
We use microRNA microarrays from LC Sciences. They have a good reputation in the field, although the Exiqon LNA arrays are also excellent. We have found the best strategy is to look at an overall expression profile of the 1,000+ miRNA genes presently identified in the human genome using a small cohort of patient samples via miRNA microarray, and then to verify the results in a larger patient sample population using an independent technique such as quantitative RT-PCR.
One of the questions we grapple with is: how many human samples for each clinical group is enough to determine a real trend or correlation of miRNA expression with a specific disease state for our initial miRNA microarray screen? It really boils down to the economics question: how many microarrays can you afford, taking into account that you will need to include five patient specimens per sample group to provide a good minimum range?
Regarding the confirmatory techniques, we use the stem-loop reverse transcription TaqMan-based real-time PCR assays (ABI, miRNA Taqman assays). These provide high accuracy and sensitivity, and although they are expensive, they are superior over the SYBR green fluorescent reporter method.
Another high-throughput miRNA profiling technique utilized in the laboratory is the microfluidic miRNA TaqMan-based qRT-PCR low density array from Applied Biosystems. We have used this platform with good results, although I must stress the necessity to do arrays at least in duplicate, as we find reproducibility is an issue. So we mainly look at trends of expression in duplicates over reproducibility of actual fold-differences. One needs to verify the results using single-plex quantitative real-time PCR since we find some profiles will not reproduce in small scale reactions.
Finally—and we don’t use this approach—but deep sequencing (using the Illumina, SOLID, or 454 Genome Sequencer™ system for ultra-high-throughput DNA sequencing) is a powerful method to discover new miRNAs and determine if lead or star transcripts are differentially expressed and possibly important in a given cancer group. This technique is expensive, but it is getting cheaper every day. The main issue with the deep sequencing technologies, however, is the quality of the bioinformatics analysis, since this approach requires huge datasets.
We use both microarray profiling and deep sequencing. Microarray because it is economical and the turn-around time is fast and allows the validation of a targeted group of sequences that are identified using deep sequencing; deep sequencing because it gives us much more information and, most importantly, it can lead to the discovery of novel small RNAs, including microRNAs. So far we have used the Roche 454, Illumina, and two different Helicos single-molecule sequencing platforms (tSMS and Direct RNA sequencing).
We have used both microarrays and deep sequencing for miRNA profiling, with microarray serving as a readily available technique, and deep sequencing providing additional information about miRNA than that which is already known.
It depends on whether you are doing an initial study, or if you have already identified your target. We use qRT-PCR TaqMan assays for the latter and we have done microarray studies, using the Affymetrix® microarray for the former; these are our main tools. We previously used Combimatrix chips (micrometric semiconductors that drive in situ synthesis of oligonucleotide probes), but that was a while ago. We then switched to the Affy product. It is convenient to have multiple platforms, as no single one is perfect, so we double-check and cross-validate everything. Usually, what we see with the arrays is pretty much what we detect by qRT-PCR.
Blockmirs are steric antisense blockers that bind to specific microRNA binding sites in target RNAs to prevent microRNA binding to the same site. Although the blockmir concept is new, it offers therapeutic possibilities. Are you working with this concept and, if so, how do you fit it into your program?
Currently we are not working on this concept. However, like any other payload in the field of cancer therapy, blockmirs require a tumor-targeted delivery system for them to work effectively.
This is a really interesting concept and it solves a roadblock in miRNA-based technologies and therapeutics, that is, how to regulate a single miRNA target when these small RNAs potentially control upwards of 100 targets. However, we have not tried this technology in the laboratory.
One issue that might be problematic with this type of technology is the issue of miRNA combinatorial regulation of a single mRNA target. Some mRNA 3'UTRs are predicted to have 10s to 100s of miRNA binding sites. It is hard to predict which ones are actually important in vivo, and how many unique blockmirs or target protectors are needed on a single 3'UTR to control expression.
We are not working with blockmirs right now. But in principle the concept of blockmirs can be used in our experiments to interrupt specific miRNA:mRNA interaction, as compared to traditional blockers that block all miRNA:target interaction.
Yes it does work and we have used it to verify the function of a miRNA.
We have used target protectors called Vivo-Morpholinos, made by Gene Tools. They do the same thing as blockmirs. They validate our findings, allowing us to look at one gene at a time instead of a whole set of genes targeted by a particular miRNA. Delivery is not straightforward, as they were originally developed for transduction in fish. Morpholinos can be transfected into animal cells using electroporation but you have to use so much that this raises concerns over whether your findings are physiologically relevant.
I do have a lot of interest in the blockmir approach. I don’t know if the technology is really there yet, but it appears that it might be a good way of avoiding the side effects of targeting the microRNA.
One of the main problems in the microRNA field is to figure out which mRNAs are regulated by which microRNAs. What techniques do you rely upon to address this issue in your investigations?
At this moment, we are focusing on the use of well-described microRNAs as antitumor agents. Our research effort is focused upon tumor-specific targeted delivery of microRNAs, achieving a large percent of the injected dose in the tumor, and the generation of a significant antitumor effect.
We use bioinformatics programs. miRGen is very helpful in this regard, as it intersects data for a handful of algorithms. In the past, in collaboration with Asuragen (Johnson et al 2007 Cancer Res 67:7713), we used microarray studies with great success.
The powerful guiding framework in this instance was to do a time course of expression profiles following miRNA administration (let-7) to cell lines and isolating RNA 4, 8, 16, 24, 36, 48, 72, and 128 hours after transfection. This was powerful enough to identify genes repressed as a direct effect of miRNA administration versus a secondary or tertiary response.
Also in our lab, we have turned toward the simple animal model C. elegans to determine targets for specific miRNA families, namely lin-4 and let-7 homologues. This is a great system to study miRNA function in the living animal using a range of genetic and molecular tools at our disposal. Currently, we employ RNAi suppressor screens to find genes that, when knocked down by RNAi, are able to block or lessen the phenotypes observed in animals carrying deletion mutations for certain miRNA genes. We have done a candidate approach in which we test likely targets in RNAi screens identified via bioinformatic predictions, but we will also take a more unbiased approach and do a genome-wide RNAi suppression screen.
My group has primarily used in-house tools along with miRanda, a program that I co-developed while I was at the Memorial Sloan Kettering Cancer Center and then design experiments to confirm the regulation in the cell. We are also beginning to use miRNA/protein pull-down to identify the precise miRNA-mRNA interactions, with the idea that miRNAs and their targets may form complexes with other RNA-binding proteins.
We have used several in silico programs to fish out a candidate list, and we have also used expression profiling of mRNA after miRNA treatment, which turned out to be fairly informative. However, verification has to be done by mutating the actual site that contributes to such interactions between the miRNA and target mRNA.
Once putative miRNA targets have been identified by microarray analysis, one approach is to simply seek confirmation using classical luciferase-based assays, but that will only yield limited information: that the specific tested sequence is being targeted by certain microRNA. It doesn’t provide information about the regulation of the endogenous gene.
The approach that we are using involves the mutation of the targeted site at the 3'UTR in the context of the complete cDNA of the gene of interest, and then we put this mutated version back into the cell. We believe this to be a better way to analyze the real impact that post-transcriptional regulation of a gene by a miRNA has in the cellular context.
This issue arises once you have established a direct target. So far we haven’t done any HITS CLIP or other co-immunoprecipitation-based experiments, although there certainly is a value in those direct approaches. We have collaborated with Rosetta Inpharmatics, which was by then a Merck & Co. subsidiary. Michele Cleary’s group there carried out a screen whereby they introduced mimics into cells, and 10 hours after transfection, looked at the immediate targets of the microRNA by mRNA profiling. It’s not a perfect tool, but almost everything we’ve got from that screening has been successfully validated by other methods. At least as a first pass, mRNA profiling gives you really good clues. This is really an excellent hypothesis-generating tool.
A nanotechnology platform has been developed that can deliver therapeutic siRNA specifically to primary and metastatic cancer after systemic treatment. Is your group using or considering this approach? If so, how have you specifically used it? If not, what approaches do you favor for siRNA delivery and why?
The key requirement for the development of an effective cancer treatment is the systemic tumor-targeted delivery of the therapeutic molecules to both primary and metastatic tumors, bypassing the normal tissues. To translate such a technology into clinical applications, we have been developing a nanotechnology platform to systemically deliver siRNA, miRNA, ASODN, as well as other molecular medicines. This platform technology is a tumor-specific, nanosized immunoliposome complex that is composed of the molecular payload encapsulated by a cationic liposome, the surface of which is decorated with a targeting moiety that is an antitransferrin receptor single-chain antibody fragment.
We have used this approach to deliver fluorescently labeled siRNA as a reporter, and therapeutic siRNA/miRNA in vitro and in vivo. We have shown tumor-specific targeting (primary tumor and metastases) by systemic delivery. We have also demonstrated antitumor efficacy (tumor elimination) in a number of tumor models with minimal toxicities.
A striking feature of this platform is the high percent of the intact payload delivered specifically to the tumor tissue and retained. In numerous models with various tumor sizes, we have determined by Northern analysis the percent of the injected dose (ID) present in the tumor and normal tissues. A significant percent of the ID (as high as 40%) was reproducibly documented to be intact in the tumor, with only minute amounts detected in the normal tissues such as liver and lung.
I presume you are referring to the recent nanoworm work? No, we don’t use this or any other nanotechnology yet, but we plan to get into the field soon. For our current studies and from a convenience perspective, we use lipid-based transfection methods to delivery miRNA/siRNAs into cultured cells. It is convenient because it is widely used and hence is easy to troubleshoot, and generally gets the job done for our purposes.
Many technologies have been tested for delivery of siRNA into various cells and tissues, but there is not a universal solution. At the moment we are working on cancer of the liver so we are very interested in the nanotechnology that you mentioned.
To become a viable drug, RNAi molecules should remain stable in biological fluids, avoid rapid clearance, penetrate the target tissue, be taken up by the target cells, and, finally, activate the RNA-induced silencing complex for RNAi in the cytoplasm of these cells. Could you describe how you are tackling these challenges?
In the bloodstream, nondelivered siRNA is completely degraded in less than 10 minutes after systemic administration. In contrast, use of our aforementioned delivery system resulted in 10% intact siRNA still present in circulation up to 24 hours post-systemic administration. When encapsulated in the delivery system, the half-life of intact siRNA was found to be approximately 8–12 hours. We have also shown deep penetration of large solid tumors by the delivery system. Using this approach at the cellular level, we also observed tumor cell-specific delivery, efficient uptake of the complex into the cell, and accumulation of the payload in the cytoplasm. Down-modulation of the target molecule and downstream effectors was observed both in vitro and, more significantly, in vivo.
A number of factors contributed to being able to overcome these challenges: highly optimized, engineered formulations of the nanoparticle, presence of a targeting entity on the surface of the nanoparticle, and making use of receptor-mediated endocytosis.
We are not dealing directly with these issues in my laboratory at the moment. Fortunately, the stability of the synthetic miRNA inhibitors or mimics is not the problem since they can be chemically modified to resist degradation and enhance cellular uptake. The true barriers in the field are how to efficiently and specifically deliver the miRNA/RNAi-based drug to the target organ or cell type.
Our lab is not currently involved in developing direct strategies to tackle these therapeutic challenges of siRNAs. That said, stability and efficacy can be addressed to a great extent by nucleotide modifications (such as LNA or 2'-O-methyl). For delivery, as you noted earlier, promising nanotechnologies are on the horizon, and we look forward to using them soon.
We have so far solved how siRNA can be degraded in blood, thus making rational design and modification of siRNA for superior stability possible. We have also solved how siRNA discriminate closely related sequences. Currently we are working to understand the process of siRNA entry into the cells.
The challenge is to bypass the liver, as most nucleic acid-based compounds that you put into the circulation end up there. But if you are trying to deliver RNA therapeutics to liver metastases or liver cancers, this drawback could become an advantage. Others have investigated the possibility of delivering the drug locally, for instance to the brain, where it is virtually impossible to deliver a compound unless you inject it directly into the ventricles. I think there are really formidable challenges. You see a lot of competing claims that you should use naked RNA, nanoparticles, or viral particles. This is why we are strongly oriented toward resolving the basic scientific question of the nature of RNA silencing.