When it comes to live cell imaging, planning and preparation is the key to a successful experiment. While planning the equipment configuration is not glamorous it is fundamental to success.
First things first: Specimen type
Whether working with adherent cells, suspended cells, tissue, or artificial membranes, each specimen type can be imaged in a variety of ways depending on the type of microscope and experimental protocol. These tips will get you pointed in the right direction.
Know your scope
A scope, like an environmental control system, is made up of a base configuration and a number of optional accessories.
You may have an upright scope providing a birds eye view or an inverted scope providing a worms eye view. Each has a choice of optional optical components that define its functionality.
For upright scopes live specimens can be positioned on the scope using the following methods:
Adherent cells can be plated either on the bottom of a dish so that they are laying down and observed with either a low magnification long working distance lens or a higher resolution lens (water dipping lens or immersion) can be used providing there is a coverslip over the specimen.
If cells are to be perfused in a discrete flow channel, they should be plated on a coverslip that is incorporated into a flow cell where the cells will be in a hanging orientation with media flowing immediately under them.
Alternatively cells or tissue can be plated on the bottom surface of a fluid optical cavity where media flows over the cells and observation occurs through the coverglass above the cells, and through the media in flow.
If no flow is required, cells plated on a coverslip can be placed face down on a spacer in a dish so that media is contained between the bottom of the dish and cells on the coverslip.
Cells in suspension usually need to be contained by either a parallel barrier such as two coverslips with the specimen in between to limit the range of motion or a thixotropic media, and are typically observed with low magnification lenses.
Tissue in any form such as natural or artificial membrane can be imaged in a variety of ways. The simplest is having the tissue resting directly in the bottom of a dish within media being observed with a dipping lens. A dry lens can be used. However, can run the risk of condensation on the lower element if the specimen is warmed. A Coverslip Lid can be used as a barrier providing the objectives being used have a long enough working distance.
Another popular method is placing the tissue on a nutrient membrane surface such as a Corning Snapwell™ membrane.
The Delta T dish provides the optical, thermal and fluidic containment and the membrane provides a porous surface for nutrient exchange. [A small tube can be introduced into the side of the Snapwell to supply fresh nutrient media to maintain long term viability of the specimen.
Note: In some cases it is best to isolate and contain the atmosphere in the dish when using dipping lenses. Atmospheric Control Barrier Rings are used for that purpose.
CO2 and perfusion can be added with the appropriate supplemental adapter.
For Inverted scopes, there are more options to view live specimens because of the obvious advantage of gravity. Most cells will preferentially plate toward gravity thereby giving the objective a worms eye view of the “footprint” of the cells on a flat optical plane. There are many configurations to choose from depending on the experimental protocol. If a discrete well defined flow of media is not required for cells, the following configurations are preferred:
Adherent cells in a dish open to the atmosphere (usually short term imaging less than 20 minutes)
Adherent cells in a dish sealed from the atmosphere (longer imaging time but usually not days)
Adherent cells in a dish sealed from the atmosphere while being perfused.
(much longer imaging time, but media has to be pre-equilibrated prior to entering the dish)
Adherent cells in a 5% CO2 controlled dish without perfused
(required when CO2 dependent media is used)
Adherent cells in a 5% CO2 controlled dish perfused
(continual or intermittent perfusion is required with CO2 dependent media)
Cells in suspension usually need to be contained by either a barrier to limit their range of motion or a thixotropic media and are typically observed with low magnification lenses.
In some cases where the inherent fluorescence of media contributes to background fluorescence, a coverslip affixed to a sleeve can be lowered near the top of the specimen plane to reduce the volume of media above the specimen.
There are two ways of observing specimens in a FCS2.
Note if using fluid coupled lenses with physiologically warmed specimens it is essential to also warm the objective to prevent a 5-7 degree temperature drop at the specimen. To avoid excess heat radiating from a lens heater that can over heat the specimen by convection, heat must be applied to the objective at the most efficient location and method while measuring the heat propagation through the objective. This method measures the heat propagation inclusive of external factors and regulates to the expected temperature at the specimen plane. The most harmful technique is to use an inefficient heat transfer device that only regulates itself independent of the objective. This method does not and cannot compensate for the continual heat sinking effect of the nosepiece.
There may be times a sophisticated environmental system is not needed. For instance, when doing short term imaging. It is still a necessity to prevent the specimen from cooling down. In this instance a peripheral warmer will suffice. The most common warmer is simply a heated metal plate resting on/or in the stage so that the specimen is heated. Unfortunately, the specimen is displaced in the Z axis due to the thermal expansion of the plate, not to mention the additional drift due to the absorption of heat by the stage. This effect is now unnecessary. An advanced design to eliminate this problem is accomplished by placing the specimen in a peripherally heated structure where both the specimen plane and the support surface for the heated structure are in a common nodal plane so the heated structure is resting on a Z axis stable surface that insulates heat from the scope. The Stable Z is the only peripheral warmer to employ this technique.
Depending on protocols, it might be necessary to supply 5% CO2 to media in order to maintain the pH of the media that is being perfused through an atmospherically closed chamber or dish. This can be easily facilitated with the following technique:
Start with a 5% CO2 tank equipped with a demand regulator (a demand regulator converts the tank pressure to ambient pressure). This is a great first step because now it is simple to move the gas from the regulator to water (or media) by virtue of its volume using a simple, inexpensive peristaltic pump. To adjust the flow rate, set the pump to deliver one bubble every 3 – 10 seconds. Each bubble is about 15 microliter. Then multiply the bubbles per minute by 0.015. This is a lot easier and less expensive than a precision pressure regulator.
Note: make sure the media is pre-equilibrated in a CO2 incubator before starting this process. Keep in mind that the purpose of this configuration is to maintain pH not establish it.
Two protocols that achieve these results:
- 5% CO2 is bubbled directly into media that will be transferred to a closed system cell chamber.
When providing gas to an atmospherically regulated environment, such as a lidded dish, the incoming gas must be 100% humidified. This prevents a shift in osmolarity by displacing the 100% humidity already in the dish with a dryer incoming gas. It is necessary to set up the CO2 delivery as above but deliver the gas to a micro-humidifier before the gas goes to its destination. The Bioptechs Micro-Humidifier bubbles the incoming gas through heated water in a closed tube like cylinder. The air or gas space above the water becomes saturated with water vapor. Finally, the outflow gas is captured in this saturated airspace and transferred back through the heated water and short coupled to its destination so that the vapor does not have a chance to condense. 10ml of water can last up to 2 weeks.
- 5% CO2 at 100% humidity must be the levels that arrive at the enclosed airspace above the media in the dish where cells are plate.
? My PI just informed me our new project is to investigate the instantaneous effect of a compound on live cells. Some of the setup is intuitive but now I have run into issues I didn’t realize would be problematic. For instance in order to observe my cells what kind of containment is best to use? How do I expose the cells to an alternate media without compromising them? How do I precondition my media? How long does it take to acquire the expected changes? Who can help?
! Fortunately, you are not the first person to encounter these issues. As it turns out, this is a routine process used to investigate many cellular reactions. One by one in order, the type of containment vessel should be a parallel plate flow cell having a coverslip surface for imaging. There are several commercial ones available. The one most recommended is the Bioptechs FCS2.
When comparing to the others, the FCS2 was the only one that allows you to select or create any flow geometry you want including laminar flow instead of being rigidly confined to the manufacturers design. It also provided temperature support uniformly over the entire aperture of the chamber. It is compatible with all modes of microscopy you may intend to use and is easy to assemble. When it comes to perfusion you may be thinking it is a plumber’s nightmare but in fact it is already worked out. If you are working with two sources of media, one is just a nutrient media and the other has the variant factor. The plan is to image cells on the scope for about 30 minutes with nutrient media to demonstrate they are “happy” and not compromised in their foreign environment. Then, depending on experimental protocol, introduce the alternate media containing your variant factor while acquiring images of the effects.
If the expected reaction takes place slowly you can manually introduce the variant, but if the protocol requires a rapid introduction of the variant that is where it gets a little tricky and is best to automate the process. There are two sources of media coming together external to the optical cavity where the cells are plated. Therefore, there is a dead-volume between the adjoining flows and the cells, and a small diffusion gradient that occurs when one flow follows another. The goal is to get the variant factor to the cells at its source concentration ASAP to record the immediate reaction. To accomplish this there is a computer program that controls the flow rate of two independent, smooth flow, peristaltic perfusion pumps. The program allows you to; establish a maintenance flow of nutrient media for a predetermined amount of time that you control, then ramp down the nutrient flow while accelerating the flow of the variant factor, then sustain that rate for a for another determined period of time you chose to ensure the concentration of the variant media is at its peak. The Variant flow rate is then de-accelerated to a minimum rate for imaging. The process can then repeated to rinse the variant factor and restore the prior environmental conditions. This saves a lot of headache
making these transitions especially when it has to be consistently repeated many times for statistical reasons
If you can’t get away with hepes media, you will need to sustain a CO2 tension in the media to maintain proper pH. A number of CO2 systems are available but the one that makes the most sense starts with a 5% CO2 and air tank under pressure then reduces the pressure to ambient with a regulator. That way the gas mixture can be metered out by virtue of its volume with a simple peristaltic pump. If you place a 14 gauge tube into a flask containing media each bubble that rises is about 15 micro-liter. Therefore, all you have to do to establish the flow rate is count the bubbles per minute and do a little arithmetic!
The amount of time it takes to acquire results will depend on the concentration of the variant factor and your cells.
Cozy Cells – Insight about Humidity and Ph Comfort
As the holidays are rapidly approaching many of us are retreating to enjoy our warm homes with loved ones, hot chocolate, some good books, and lots of delicious food. Well that’s the fantasy anyway.
For lidded single-well dishes or closely enclosed multi-well time lapse plates at physiological temperatures, the airspace to culture medium ratio is low. Therefore, the RH in the airspace above the media will reach 100% RH almost immediately. Once humidity is at 100% water in the media will no longer evaporate, therefore, ideally when perfusing a specialized gas mixture or the more commonly used 5% CO2 in air to the existing 100% RH airspace in the chamber, the inflow should be also be entering the chamber at 100%RH. Unfortunately, like the fantasy holiday that is often not the case.
The problem is that if gasses are entering the humidified chamber at less than 100% RH it causes the water and carbon dioxide in the media to evaporate to maintain the humidity levels which effectively dilutes the media and causes osmolarity changes within the dish that can in the end affect cell phenotype. Likewise, flow rate is also important to prevent Ph shifts from occurring. As a rule of thumb, the volumetric gas exchange rate should be on the order of 1 to 5 volumes per hour to prevent pH or osmolarity shifts. Now, how to keep the inflow of gasses at 100% RH is the question.
Creating 100% RH inflow can be accomplished with a well-designed humidifier that directs gas through warmed water in a manner that the humidified gas is captured and transferred to the dish. However, even though the house and the car are warm, it doesn’t prevent us from getting cold walking down the drive to get in the car. Nor does it prevent our gasses from condensing to water and 60% RH in the tubing when traveling from the warm humidifier to the cozy chamber. Much like keeping the car in a heated garage attached to the house, if the 100% RH outflow gasses are captured and in the vessel and passed back through the heated base then close-coupled to the cell chamber it prevents the humidified gas from condensing before it reaches the enclosed cell chamber.
Using a quality humidifier in combination with a demand regulator and a low flow peristaltic pump to meter flow rates creates an easy and cost effective way to establish the correct gas and RH conditions for live cell microscopy. While we may have to endure in-laws, shopping, weather, and chaos we can at least take solace in the peace of mind that our cells are as happy as we are to ring in 2017.
Tissue observation in a snap!
Imaging isolated live tissue without it drying out
Just because the specimen is in focus now, doesn’t mean the rest of the time lapse images taken are depicting healthy tissue. The means of environmental control can make the difference between the acquisition of quality data and the acquisition of frustration. The following approach can make tissue imaging a matter of a simple routine practice instead of an annoyance.
When it comes to imaging tissue in vitro on an upright scope some intuitive techniques go a long way to make the job easier. In the planning stage, the physiological requirements of the specimen are already known, but consideration for the characteristics of the microscope objective must also be taken into account. A commonly preferred objective for imaging tissue is a water dipping lens because it provides a high N.A. image. However, either a standard dry or even a coverslip coupled lens can be used with the appropriate technique. In all cases whether the specimen is mammalian (needing temperature control) or not, the challenge is to keep the specimen mechanically positioned and supported with nutrient media; To provide nourishment, removing waste products, and to keep it from drying out without interfering with any limitations imposed by the optics.
A favorite technique is to place the tissue on a Corning Snapwell™ with the membrane facing up and resting in a Bioptechs Delta T Culture Dish. Using this configuration media can be applied to the 3mm space under the Snapwell™ membrane so that a tissue resting on top of the membrane can exchange nutrients and waste products throug
h the membrane. The Delta T has exactly the right inner diameter to accept the Snapwell™ and is available with or without intrinsic temperature control. It is important to note that if temperature control is needed, the Delta T will warm the specimen without having to heat the stage. This is a big advantage because if the stage must be heated, you end up inducing an undesirable Z axis drift. Another advantage of the Delta T Dish is that the stage adapter for Delta T Dishes is already machined to accept two perfusion supports. These supports enable the introduction of fresh media under the membrane surface of the Snapwell™ and remove excess media on the other side of the dish, thereby preventing overflow. This configuration can be used with dipping lenses, dry lenses, or if your lens requires a coverslip, a coverslip can be placed on the tissue. Continual perfusion is easily controlled using the following configuration: A peristaltic pump (It is highly recommended to use an analog, DC motor driven, tachometer regulated pump that is capable of retaining two different sized tubing in the same pump head). The smaller diameter tubing is used to supply media to the dish and the larger diameter tubing is used to aspirate the dish. In this configuration the media is always removed faster than it is coming in, thus preventing overflow. By simply adjusting the height of the pickup needle the fluid level in the dish is established in such a way that there will be no chance of over aspirating.
Bonus tip: If the specimen is warmed above ambient it will be necessary to also use an objective heater for two reasons. If using a dipping objective, it would be an adverse effect for the temperature of the objective to affect the temperature of the specimen likewise, if using a dry objective it is important to prevent the formation of condensation on the lower element of the objective.
Sheerly Brilliant Affinity Studies
Affinity studies are riddled with variant factors that can create an array of frustrating, irregular data. As if calculating the dyn/cm² on sheer force experiments isn’t fun enough, the unfortunate reality is that when using a radial flow cell those sheer forces cannot be adjusted to fit varying protocols. However, by using a chamber that is designed to enable customized flow characteristics a broader array of research options exist. The advantages of using a customizable flow cell system over a traditional radial flow cell range from the intrinsic micro-environmental control capabilities to increased speed and accuracy. Traditional radial flow cells are clumsy and limited to fixed flow characteristics that consume large volumes of expensive media and/or cells. In contrast a customizable flow cell chamber is a more efficient means of evaluating adherence characteristics within sheer forces.
Flow cell chambers like the exceedingly popular FCS2 and FCS3 System are closed micro-observation chamber systems that offer uniform temperature control and is fully compatible with all modes of microscopy for time lapse imaging. Since these systems are designed to handle high and low volume perfusion rates through the use of a micro aqueduct perfusion technique it is easy to do customized flow profiling to characterize the sheer forces within the flow range.
- Electrical Enclosure, Temperature sensor, Heater contacts (Can also be detached to sterilize the perfusion tubes)
- Upper Half (facilitates perfusion)
- Perfusion Tubes (14 gauge)
- Upper Gasket (seals micro aqueduct slide)
- Microaqueduct Slide (An optical component which integrates constant or variable flow perfusion, temperature control via an electronically conductive coating, and enables Koehler Illumination)
- Singular lower gasket (customizable internal geometry to define flow characteristics for sheer force experiments)
- 40mm #1.5 coverslip (Surface where cells are plated)
- Self locking base (Designed to assure parallel uniform closure, eliminate leaks, & broken coverslips)
A fluid pathway is formed by separating the Microaqueduct Slide from the coverslip with a single silicone gasket of a determined thickness. By changing the geometry of the lower gasket the flow characteristics can be converted from laminar flow to any sheer characteristics desired. Perfusion access to this flow channel is made through two 14-gauge needle stock tubes (3) on the sides of the chamber in fluid contact with the Microaqueduct Slide that comprise two “T” shaped grooves cut into the inner surface of the slide. The “T” groove allows the media to seek the path of least resistance and become nearly laminar before flowing across the cells. Alternatively, the introduction of media can emerge from a single point and diverge to a wider outflow area thereby simulating a wedge of a radial flow cell with user customizable flow characteristics. This technique also eliminates the need for a metal perfusion ring and additional gaskets, which are the limiting factors, required by most conventional chambers. (see drawing below)
There are two ways of introducing shear. One is to have a laminar flow region and vary the flow rate . The other is to have a gradient flow region and a constant flow rate. The later provides greater experimental flexibilities. For the practice of sheer force experiments the internal gasket geometry determines the flow characteristics.
Step one is to functionalize the coverslip or micro-aqueduct slide by coating it with an ECM, lipid bilayer, or other variant factors. Cells can be either plated on this surface or suspended within flow, depending on protocol.
Step three Transfer to microscope
Step four Perfuse cells or media while acquiring images
Hocus Pocus Stay in Focus
Have you been frustrated by frequent focus shifts during image acquisition? While it may sometimes seem like there are evil forces working against you, it is not magic but physics. When warming a specimen that does not require a highly sophisticated micro-environmental control system the choice tends to lean toward a warming plate or stage top incubator. These devices provide peripheral heat transfer that propagates everywhere including the specimen. However, they have a known side effect of frequent of Z axis shift that greatly effects the imaging process. This Z axis drift is a result of thermal expansion that is occurring on all surfaces that are exposed to the radiant and conductive heat being produced. All materials have a thermal expansion coefficient that indicates the degree of expansion relative to the difference in temperature. Nearly all stage top specimen heating devices are made of a metal plate that rests on the stage, but metals have fairly high thermal expansion coefficients. As the metal warms to the necessary temperature to transfer heat through the air, the plastic, and finally to the media it is simultaneously expanding not only itself but all of the other components that heat is coming in contact with including the stage. As this expansion occurs the only direction the specimen has to go is up, whereby causing it to drift out of focus. Not only is this a highly inefficient waste of energy but less productive as well. Interestingly enough, within a heated solid there is a nodal plane from which thermal expansion occurs, but not inclusive of. By placing the support surface and the dish on that nodal plane thermal expansion can occur both above and below the specimen without the specimen moving out of focus. Concurrently, it is important to note that the only heat that actually affects the specimen is the heat produced immediately adjacent to the specimen, therefore, only a small isolated footprint heating surface is all that is necessary to warm the specimen. When these improvements are combined with the use of a material that has a near zero thermal expansion coefficient for the plate the effects of thermal expansion are controlled and isolated from effecting the rest of the scope. This prevents thermal drift and wasting energy on heating every component that comes in contact with the plate. The result is a warmer that does not induce Z axis drift (see Stable Z) and of course a happier researcher.
Inverting Membranes to Enhance Research
When you think of plating cells on a membrane insert what typically comes to mind is a process by which cells are plated into the upper chamber of a membrane insert that is placed into a well. This protocol has a wide variety of uses, but also has some inherent constraints. The primary dilemma is the inability to view or image the cells on the membrane during the introduction of variant factors. There are volumes of information that can be learned from the ability to directly observe and quantify the behavior of membrane cells under different conditions. Achieving the ability to image cells on a membrane insert is a fairly simple process and can be accomplished in a couple ways.
Protocols for seeding cells on the underside of insert after cells have grown to no more than 90% confluency
1. Remove inserts and place upside down so that the inserts are resting on the upper lip with the membrane facing up
2. Place sterile piece of tubing over upside down membrane insert
3. Pipet cells across entire membrane
4. Incubate cells to allow them to attach
5. Remove from incubator and remove tubing from membrane insert
6. Place back into well in normal orientation
7. Apply media to wells
1. Turn the plate and inserts upside down so that the inserts are resting on the upper lip with the membrane facing up
2. Pipet cells on to the membrane in the form of a large drop so that surface tension is maintaining the shape of the sample
3. Incubate cells to allow them to attach
4. Remove from incubator and place plate over inserts and turn back over to normal orientation
5. Apply media to wells
After cells are seeded on the underside of membrane insert, for optimum imaging, the insert should be placed into a microenvironmental control system to simulate host conditions for time lapse images to be acquired (for ideal imaging check out the Delt T System). By simulating host conditions it is possible to directly image cells on a membrane with the ability to adjust the proximity of the membrane to the focal plane of the objective. By imaging cells on a membrane insert in a microenvironmental system; long-term time lapse of the cellular behavior and permeable activity is easily accomplished.
So you have everything planned perfectly; the lighting, media, a camera to capture the moment, and it all goes cold! What went wrong? Cells in vitro tend to be very moody about their environment and nothing kills the mood quicker than getting cold. Two important elements of imaging with high numeric aperture lenses that need addressed are, first the method of which the specimen being heated and secondly the heat sink factor of the objective.
The biggest problem with using peripheral heating systems such as stage heaters and stage top incubators is that it relies on the heat from the base to radiate to the specimen when a large percentage of that is absorbed by the microscope. Unsurprisingly, this causes a temperature variant across the specimen plane that heavily effects the cells growth and behavior. With such an inefficient heat transfer the re-equilibration time with changes in temperature or media introduction tends to be very slow. The thermograph below on the right indicates the disadvantage of peripheral heating. This is a thermal image of a 50mm culture dish in the center of a 100mm diameter uniformly heated, 3mm thick, aluminum plate with a 25 mm hole in the center. This image was acquired after 20 minutes of equilibration. Note the high temperatures of nearly 60° C, that it takes to reach 37° C in the specimen area. In this case heat that is not beneficial to the specimen is sunk into the stage causing Z-axis instability. In contrast the thermograph on the left shows the efficiency, accuracy and uniformity of a heating system that directs efficient heat to only the specimen (Delta T™ system). Notice the temperature of the stage adapter. It is nearly the same temperature as the room temperature background. The dotted oval shows where the edge of the stage adapter is in visible light. Only the specimen and media are heated. Power consumption is 0.9 watts because heat is only applied to the specimen area. There is no heat transmitted to the stage.
Once the specimen is being heated accurately there still remains the problem of the objective acting as a heat sink. In this instance the optical coupling medium (oil, glycerin or water) acts as a thermal coupling medium and draws heat away from the specimen. The thermal mass of a fluid coupled objective is overwhelming when compared to the thermal mass of the cells. To eliminate this thermal gradient, it is important to accurately and carefully warm your objective (Objective Heater). It may also be necessary to isolate the objective from the nosepiece turret for proper objective temperature regulation with a Thermal Spacer. When selecting an objective heater it is essential to select one that is referencing the temperature on the focal plane of the objective since this is the point that can affect the specimen. By using a system that is specifically designed to slowly heat the objective then hold the objective at the set point value you have eliminated the cooling factors that prevent cells from their natural behavior all while protecting your objective from damage of overshooting and inefficient heating.
Speeding up Cell Plating for Imaging
Sample preparation has always been a notoriously time consuming task that tends to detract from the more essential functions of collecting and analyzing data. A few key factors come in to play with improving the process of plating cells for imaging and the time that it takes. Cell adhesion, media re-equilibration, an unobstructed path for free migration of cells are some of the important factors to improve plating efficiency. Cell adhesion is pendant on the surface the cells are being plated on. The chemical composition of the glass affects cell adhesion and all glass surfaces are not created equal. It is best to use glass that is alkaline free and designed for cell adhesion (check out the Bioptechs Delta T Culture Dishes, FCS2 coverslips, 30mm ICD coverslips, and Microaquaduct slides). Sometimes an ECM is required depending on the cell type and protocol, however, in all cases cell plating is improved with the use of Culture Cylinders. A unique attribute of using a Culture Cylinder for plating is the rapid, organic adhesion in a defined location on the dish. This ensures the best cells are in the most viewable position on your substrate for imaging. Culture Cylinders have made innovative improvements over pouring tripsinized cells into an entire dish of media and waiting. Loosely related to cloning rings; Culture Cylinders in contrast are autoclaveable borosilicate glass polished on one surface optically flat to create a hydrostatic seal to the substrate that cells are plated on so that grease is not required. This minimizes the volume of media used for cells to re-equilibrate to once tripsinized, thereby allowing cells to plate faster. Also by eliminating the use of grease there is no contamination induced by plating cells or obstructive surface to inhibit cell migration. As noted in a manuscript by S. Mathupala and A. Sloan (2009) using grease with cloning rings has its own set of inherent problems that have a clear effect on the plating process. By applying this method, cell plating has become more efficient and also enables a variety of new experimental configurations with multiple Culture Cylinders
Mathupala, S. P., & Sloan, A. E. (2009, April). An agarose-based cloning-ring anchoring method for isolation of viable cell clones. Retrieved August 19, 2016, from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2727865/