Three letters have been very busy in 2020: P. C. and R. The Polymerase Chain Reaction. It’s a tiny, thermally controlled, cyclical, enzyme-driven, chemical reaction which lets scientists identify the presence of DNA both very specifically (you identify what you aim to, and not what you don’t) and very sensitively (you can identify amounts too small to find otherwise). While both benefits have featured in many debates this year, I haven’t spoken much about the type of PCR used in the pathology lab; real-time PCR (
The PCR was first described in a crude form in 1987. Improvements to PCR product detection afforded by the addition of new detection methods closed the lid on otherwise damaging PCR-product-carryover contamination issues. We”’ come back to this.
In 1991, Holland and team described the foundation for closed-tube, real-time PCR. They made use of two key features of a DNA-dependent DNA polymerase (DDDP) enzyme derived from the heat-loving Thermus aquaticus, (called “Taq DNA polymerase”) bacteria: its ability to function at high temperatures (>90°C) and its ability to chew up DNA that was in its path.
I’ve talked in some detail about PCR before – The mechanics of the polymerase chain reaction (PCR)…a primer – but I’m going to cover a lot of it again here as well.
The detection of the DNA PCR products moved PCR from its conventional form to its real-time (rPCR) form. This is the part that really distinguishes the type of PCR conducted by a microbiology pathology laboratory today, from what was used pre-2009.
Once we’ve collected a sample and have the details recorded, we transport that to the lab. In the lab, we purify away all the snotty mucous and cells and breakfast, and we’re left with just the nucleic acids – ideally just the RNA for RNA viruses, and the DNA for DNA viruses.
SIDENOTE: We most often collect a sample to test when someone is symptomatic – when they have symptoms of an illness.
Sometimes we’ll screen well people for research or because we’re looking for presymptomatic infection or asymptomatic infection. The point of a lab test like PCR is to find something that isn’t normally part of your body.
The presence of the RNA virus your Doctor asked the lab to look for signals a likely cause of the illness you have.
While the virus can be present without you being symptomatic, more often than not, we don’t test you if you’re well.
PCR is about amplifying DNA – so how does it handle RNA viruses?
After the extraction process and if we’re hunting for the presence of an RNA virus, we first have to make DNA out of the virus’s RNA genome; the PCR part works most effectively on a DNA template, not an RNA one. This happens through a process called reverse transcription (RT, not to be confused with “real-time”).
Reminder about DNA and RNA
Deoxyribonucleic acid (DNA) is the code of life in the form of a string of just four nucleotides. Four nucleotides make every biological entity into what it is. How cool is that? The unique sequence of the nucleotides in these strings of code defines each biological entity on the planet; from their toenails to their spike proteins to the stripes in their fur. The full sequence of all their genes is called their genome.
A lot of viruses that infect humans use RNA not DNA for their genome; coronaviruses, rhinoviruses, influenza viruses, respiratory syncytial viruses and so on. This lets them shortcut some steps in making more of themselves during the repurposing of the host cell they’ve just hijacked.
RNA has a slightly different chemical structure to DNA (a ribose instead of deoxyribose sugar and it incorporates uracil in place of DNA’s thymine) and it’s usually single-stranded, not double-stranded like the DNA in our cells.
RNA is used as a template to make more RNA and the proteins that biological entities need in the form of enzymes, antibodies, molecular channels, hair, nails etc.
The PCR amplified up DNA but can also be used to detect RNA – but not directly. This is (commonly) done with the addition of an extra first step called reverse transcription (RT) which creates a DNA copy of the RNA. A PCR positive tells us the target’s genetic material is there, but it doesn’t tell us whether infectious material is…or is not…present. We know from work in the lab that PCR almost always detects infectious pathogens when present in a sample. PCR can also detect RNA and DNA when infectious pathogens cannot be cultured using cell culture systems.
The reverse transcription (RT)step
The RT step involves a pre-PCR incubation at 50°C to 60°C for 5 to 30 minutes in the presence of RNA, buffer, nucleotides and a reverse transcriptase enzyme (an RNA-dependent DNA polymerase or RDDP).
Compared with research or academic labs, higher throughput (processing and testing many specimens per day) pathology labs use kits that let us do both RT and PCR in a single tube. I’ve talked about kits and reagents and these earlier steps before if you’d like to know more.
SIDENOTE: When we look at a written-down version an RNA sequence, we read it from left-to-right which we describe as in a 5’→3′ or “five prime to three prime”, direction. Same with a DNA sequence.
But as there are two complementary strands to a DNA sequence or PCR product, the second or reverse strand has to be written in the reverse complement – backwards plus changing each nucleotide to its complementary nucleotide – to be read 5’→3′.
The reverse transcriptase enzyme – technobabble follows
The RT enzyme hops onto the right-hand or 3′ end of the RNA template, where the primer (maybe virus sequence-specific, a random sequence or a string of Ts) has bound and makes a DNA copy of the RNA template moving from right to left (3’→5′ of the RNA template strand, but creating new DNA in a 5’→3′ direction; this won’t be in the exam). So it reads as back-to-front, or in th
The new sequence is the complement of the RNA template sequence, as well as being in reverse order. What does all that mean? Let’s make up a short RNA sequence as an example:
5′ GAGGUCAUGG 3′.
During RT, wherever there is a ‘G’, there will be added a complementary “C” added to the newly growing cDNA chain; for every “U” a complementary ‘A’, for every ‘A’ an ‘T’ and
5′ CCATGACCTC 3′
…as a result of the RT step. It’s the reverse-complement of the original sequence (try reading it back-to-front and decoding it for its matching RNA nucleotides. You’ll end up back at the first sequence!).
The newly created DNA copy of an RNA template is called complementary DNA (cDNA). Now we can conduct PCR using this DNA. We call the entire process an RT-PCR. There’s more on what RT-PCR is over at Reverse transcription-polymerase chain reaction (RT-PCR)…a primer for virus detection. After the RT step, we kill off the RT enzyme using a heating step.
Detecting the products of the PCR
Detection is really where real-time PCR (rPCR) shines (yes I did). So let’s look at two different points on the evolutionary timeline of PCR’s growth into a safe and effective tool for large scale virus testing as it moved from conventional PCR to rPCR.
Names and definitions
The combinations of PCR and non-fluorescent amplicon detection assays that are run to the maximum number of cycles (the endpoint) before the tubes are opened for a detection step, I’ll call “conventional” PCR here. Those methods which use a fluorescent probe to detect the growing number of new PCR product molecules as they accumulate without need to open a tube or stop the PCR cycling I’m calling “real-time PCR” (rPCR, or RT-rPCR for reverse transcriptase real-time PCR).
SIDENOTE: Some of the anti-PCR chatter which abounds in 2020 may simply be the result of confusion about the existence of conventional PCR which runs to the endpoint before a result is recorded, and rPCR which runs to the endpoint, but which produces its results almost always before that endpoint. It’s also worth noting that at the endpoint of a PCR, there may be no relationship between the initial template and final PCR product concentrations.
There is also the non-probe fluorescent methods of rPCR using dyes (also labelled primers) which interact with the double-stranded DNA PCR products. They aren’t always as specific as probe-based rPCR and I prefer the use of a probe. Lastly, within the category of rPCR, you can have approaches that require the probe to be destroyed to produce a signal (destructive chemistries), and some that remain intact (non-destructive chemistries).
Detecting PCR product in the dark days
Conventional PCR product detection methods included agarose gel electrophoresis (discussed below), Southern blot hybridisation (often with large probes) and ELISA-like methods, using radioactivity, chemiluminescence or visible colour-producing chemistries to detect the PCR product at the endpoint.
Electrophoresis of a portion of a PCR product through an agarose gel immersed in a buffer containing a DNA stain was how we used to do things. This separates the different sizes of DNA, with the smallest moving fastest through the gel under an electric current.
Click on image to enlarge.
The PCR might have taken two or so hours back then, and the gel another 30-60 more minutes (depending on how much voltage you liked to apply 😉-and if you forgot, you’d run your small products off the end of the gel into the buffer😣. I have heard.).
Results took longer to get, plus you were most often using the size of the product to identify a positive test result. That could be problematic if it varied or if there was non-specific amplification yielding a similarly sized product or affecting the overall sensitivity of the amplification by sucking up the primers and probes and enzyme to make other products.
Methods to probe these PCR products after transferring them to a membrane or plate made the detection more specific but still required the PCR product tube to be opened, increasing the contamination risk.[2,3]
Real-time PCR added that extra layer of sequence-specific confirmatory power and produced a machine-measurable signal (radioactive, colour or fluorescent) only when the specific PCR product was produced, without needing the tube to be opened.
Let’s talk about PCR in real-time
So we know what PCR is, we know what RT-PCR is, we understand that conventional PCR used detection methods that were slower, required the PCR product tube to be opened and could involve use of toxic chemicals.
Enter real-time PCR.
In real-time PCR, primers get a new friend, the oligonucleotide probe
The addition of a non-radioactively labelled oligonucleotide probe was a breakthrough; it made the PCR process work better, provided a clearer result and was faster with fewer risks.
SIDENOTE: Before we get into the PCR cycles, I just want to mention a small issue – we only have a single cDNA copy made from the single strand of the virus’s RNA genome. To make things simpler, I’m picking up the PCR at Cycle 2 below. At this point, we will have two DNA strands (one cDNA and one new copy) and I think this makes the maths a bit easier.
The steps in a real-time PCR cycle
I’m going to use these symbols to identify the four DNA and four RNA nucleotides, to make it easier to show them building up each cycle.
As I’ve laid out over on the mechanics of PCR blog, the PCR part (we talked RT above) of the RT-PCR involves multiple steps, which together comprise a cycle (see the image in the Background section above for a graph of the temperatures in one PCR cycle). Multiple cycles comprise a PCR. The last cycle is called the endpoint. What cycle within that PCR a positive result occurs, is called its threshold cycle (CT).
The first step in an rPCR cycle is to make all the DNA single-stranded so that the primers and probes can gain access to the sequence and bind to the sequence-specific reverse complement target. This is done using heat (over 90°C) in our temperature programmable thermal cycler (an all-in-one oven and fridge, but don’t store your milk in this fridge!).
Below I’ve shown a simplified mix of the extracted human (mostly blue) and viral (for this blog piece; red, blue, green and yellow) nucleic acid strands we purified from the patient’s swab/sputum/bronchial wash/saliva/tissue/faeces/blood. Also present is one forward primer, one reverse primer and a probe labelled with two special molecules; the fluorophore (a molecule that generates fluorescence when excited by light shone at the tube from the instrument) and a quencher (a molecule that absorbs the fluorophore’s emitted energy and either disperses it as heat or light of another wavelength that we don’t care about.
Then we lower the temperature to allow the oligos to bind, or anneal, to their target. We design the probe so that it will likely bind first, as the temperature drops to something closer to 50-60°C, when the primers anneal. This all happens very quickly. There’s some wiggle room in the temperature we choose here as these are predicted temperatures which can be affected by a few things we won’t get into here (I have a bit of this info over on an unfinished page here).
In the image above, I’ve added in more of the primers and probes, and the free nucleotide “fuel” (you can’t make new strands without the building blocks) just to highlight that there are a lot of complex molecules whizzing around in that PCR tube under the influence of heat and hungry enzymes. It’s an environment that a tiny version of The Flash would do well in!
SIDENOTE: Neither the primers nor the probe will bind to the DNA if their target sequence is absent, or too different (they can often handle a few “mismatches” but not many). So there will be no new strands made and no fluorescence produced if the target isn’t present in those nucleic acids we extracted from the patient’s specimens to begin with.
Extension; the final step in the a PCR cycle
Next, we raise the heat to ~60-75°C which is the temperature at which the PCR workhorse enzyme, the heat-stable DDDP enzyme, starts adding new nucleotides to the growing strand as the copy is made. So far this is all PCR 101 stuff.
SIDENOTE: We don’t have to have a separate annealing step – with a good primer and probe design and RT-rPCR kit choice, we can instead have a single combined step that runs at ~60°C and allows for both oligo annealing and enzyme extension of the new strand with probe destruction (see next section)
As the enzyme on one of the strands travels along, it bumps into, displaces and chews up the already-bound probe. That destruction sets the fluorophore free and it can, at last, get away from the glass-half-empty negativity of the quencher; fluorescence is emitted.
That single destroyed probe and its released fluorophore (green glowy dot) aren’t noticed by today’s fluorescence detection equipment so we won’t be able to see a result for a few more cycles.
Now that we’ve done all of that, we’ve completed Cycle number 2.
In Cycle number 3 the same steps are repeated. The difference is that we start off with the two cDNA strands we had in Cycle number 2, plus we now have the two new strands we just made. We’ve doubled the number of potential target molecules.
Those with a keen eye will notice that the two new copies are a bit shorter than the original two cDNA strands. That’s because the copies only commenced from where the DDDP hopped on – which is where the primer was bound – go back up and check that out.
At the end of Cycle number 3, we have eight strands – the two original, the two made in Cycle one and four more made in Cycle 2. We’ve gone from 2, to 4 to 8 copies.
This exponential growth will continue. Let’s say we have 40 cycles. In a perfect reaction that would result in a yield of 1,000,000,000,000 (1 trillion) copies. Now imagine that instead of starting with one copy of double-stranded cDNA, we have 10,000 copies. The numbers get big fast!
The more RNA we start with the faster those DNA copies mount up and the sooner we see the fluorescence become detectable. Ironically, this means that the more RNA initially present, the smaller the CT number will be. The less present, the higher the number will be. Make sense?
Cycle number 4 is the same process – with fluorescence accumulating as we go.
How an rPCR looks in the lab
Before we get into that – let’s use a stylised set of example rPCR fluorescence data curves I just manufactured, to highlight important features.
An rPCR curve is a graphical plot of the fluorescence generated each cycle versus the cycle number it was collected on. A “good curve” is sigmoidal or S-shaped, has a good height compared to the known positive sample/control included in each run and doesn’t drift slowly upwards towards the end of the run (see the green curve). Such “late and low” positives may be dismissed or maybe further investigated to ensure nucleotide mutations haven’t started to negatively impact the PCR producing a poor result (which may still indicate virus).
You can’t tell a false positive or false negative result just by looking at the curves though. That requires knowledge of the test, the patient’s clinical picture and their epidemiological context. A PCR “diagnosis” is thus more than just a laboratory result of “virus X detected”. Diagnosis is collaborative and has important checks and balances involved; at least when the entire process isn’t overwhelmed.
Also, not all commercial versions of rPCR show a curve like this one or provide CT values.
Take a walkthrough of a real rPCR run
The first few RT-rPCR cycles (cycle number measured along the bottom or x-axis) may look like this (this is an example run I did in a past life).
This is after the RT step and the RT-enzyme killing heat step and is about 4 PCR cycles in. Boooorinng.
Some of us like to check in on our important runs every few cycles. It’s like clicking refresh on the concert ticket site! On the left vertical (y) axis we have “normalised” fluorescence. The fluorescence data are collected by the instrument at the end of the extension step of each cycle and plotted onto this graph.
You can literally watch the curves (if they appear) grow in real-time.
Normalisation for each sample is done by taking the fluorescence data from the *background* of each sample-from up until just before amplification begins (2nd derivative) and averaging. All data points for that sample are divided by that average. This is being done by the instrument’s software. this is a RotorGene instrument for those interested. Still not much to see at 10 cycles.
Jumping ahead to about 32 cycles. Hmm, what’s that pink line doing? Keep in mind that I already know this to be a positive template. It’s a synthetic (or in vitro transcribed; ivtRNA) RNA target-not virus-so this isn’t a run to detect an infection. It was actually to see how precise my pipetting of the same RNA multiple times was.
Something has definitely been happening after 30 cycles!
We can see that the fluorescence of a bunch of samples (each a pink line) has started to curve up. We could call this & go home now, but it’s better to know what else is happening with other samples & controls & review the shape & height of each curve so we let the rPCR complete its 50 cycles (we can set that final cycle number to be anything, as you’ll see, that won’t change the CT unless we set it to something dumb like 10, in which case, everything would be “not detected”).
Look at that exponential growth no! Remind you of anything else?
At over 40 cycles we can see some spacing between the curves (replicates of my ivtRNA) which suggests a little variability in my pipetting skillz, but still a pretty tight set of replicates.😎
If we take the log10 of those data (this is done using the software that drives the PCR instrument) we can see a nice long exponential (linear bit that harshly tilts upward) phase – this is where we want to set our threshold so as to capture the actively “growing” fluorescence curve. We can also ask the software to set it (if it’s the right run type and the software can do it) and it will do this by placing the threshold suitably above the noise (all that flat wobbly junk down the bottom) and in the exponential phase of the curve. The threshold (I set it here at 0.05) is important because its placement can alter the CT value by a little or a lot. In my observation, the threshold is often set or tweaked manually.
SIDENOTE: Each lab knows its PCR tests and systems, so its results are reliable. Comparing results between labs, because of aallll the variables I’ve listed here, is difficult.
What pathology labs aim to ensure they achieve is for a pathogen-positive sample to be detected and for a pathogen-negative sample to return a negative test result. Labs themselves may be accredited and have to meet standards that ensure they document and produce quality results. This can be monitored through a process of external quality control. One expert lab/company/authority sends the same panel of samples to multiple labs. They know what the detected/non detected results should be. The labs apply their protocols, having been told what pathogen this panel contains, and the labs are anonymously numbered. Later you get a general “report card” – at which time you cheer…or cry.
Within a lab, which knows its tests well, manual threshold setting can be a good thing because it calls upon the experience of those using the test who know the test’s quirks and ensures the best results are reported. In commercial kits, you may not get to see the threshold or even see the curve data – and so we rely on those systems to report a detection in the same way we would as experienced human users.
The final snapshot is of the completed run. A few things to note. I pushed the CT out to 50 total cycles (that’s a bit long because we want our results as soon as we can get them and very few clinical samples amplify anything after 40-45 cycles).
The result shows that tubes with known positive RNA in them were already clearly positive by about 18 cycles before the endpoint (their CTs ranged from to 31.44 to 32.51 ) of 50 cycles.
The result also shows that what amplified and fluoresced was what I expected to and not what I didn’t expect to. This is the case for the overwhelming majority of pathology lab PCR results whether using test made in-house or bought from a commercial kit maker.
The primers and probe bound to their intended target here and generated a signal. Ten different ivtRNA-containing samples didn’t generate a signal here – they are the flat green lines which never crossed the threshold I set (the horizontal red line). They bound different primers and probe. Nice and specific.
Also, there are four water blanks (no-template controls; NTC) in this run – do you see any black sigmoidal curves? No. That’s because the intended target is not something that just exists everywhere. In this case, it exists where I deliberately pipetted it. In the case of a virus infection, the viral RNA (or DNA) exists in your sample when you are infected, but not normally. After infection, in most instances, the virus is cleared from your body (I’m not talking about persistent or latent viral infection here of course).
The NTCs also show that there is no carry-over DNA in my mixes. My mixes and the lab environment in which they were assembled was clean. The run worked and can be reported. I don’t have a specific positive control in this run because my synthetic RNA was a positive control – it was designed to include the primer and probe binding sites targeted by the included primers and probe in the reaction.
One last sidebar: the dual-labelled probe can be labelled with different fluorescence-generating molecules (and quenchers). The one in the result above emitted its extra energy in the green wavelengths where it was recorded by the instrument.
Recall those other 10 ivtRNA replicates? They were targeted by a probe that emitted its excess energy into the orange wavelengths. And yes, they did also amplify and get detected (although I may have rushed the pipetting as they are not as tight a grouping of CTs 😣; 33.03-35.26) in the fluorescence channel. Shown in green above, theses 10 also amplified in the same PCR run, there was no leaching of fluorescence from the green channel into this channel, or cross-reactivity between probes and the NTCs were also clean at the threshold value of 0.05.
To summarise: we like real-time more than conventional PCR
The large busy high-throughput microbiology pathology lab’s biggest threat in using PCR frequently with a lot of handlers and time pressures has always been the risk that PCR product from a previous run could contaminate the general lab environment and tubes in the next run, causing a false positive.
This sort of event is a huge pain because it halts testing – sometimes for a long time – while the contamination source is eliminated. Performing the RT, the PCR and the detection in a single tube rPCR dramatically reduces the risk of this carry-over contamination.
We also like rPCR over conventional PCR because it:
- is faster
- produces a specific value that can be recorded
- doesn’t require potentially carcinogenic DNA stains (like ethidium bromide as well as others that are reportedly safer)
- adds another layer of target specificity via the sequence-specific oligoprobe
Having said all that, conventional PCR is still very much in use in research/academic labs for a variety of reasons.
If you have sincere questions that I didn’t address – please sound off in the comments below.
- Kits and reagents and viruses
- Co-detection and discrimination of six human herpesviruses by multiplex PCR-ELAHA
- Quantitative PCR-ELAHA for the Determination of Retroviral Vector Transduction Efﬁciency
- Real-Time PCR in Microbiology: From Diagnosis to Characterization
- Specific synthesis of DNA in vitro via a polymerase-catalyzed chain reaction
- Kinetic PCR Analysis: Real-time Monitoring of DNA Amplification Reactions
- Detection of specific polymerase chain reaction product by utilizing the 5′ – 3′ exonuclease activity of Thermus aquatics DNA polymerase