Linearity failure in SEC HPLC of FMDV picornavirus (TSKgel)

[mspittel]

Hi!

We are in the process of porting a Foot-and-Mouth Disease virus quantitation at 260 nm from SEC FPLC with Sephacryl S400 in an GE Äkta UPC-10 to HPLC in a TSKGel G4000PWXL column. With the low pressure Sephacryl columns we had excelent linearity over almost 2 orders of magnitude (see Vaccine 29 (2011) 7182– 7187). We bought an Agilent Infinity 1200 with a VWL detector and automatic injector as a dedicated equipment for this purpose. During early testing we had similar results as FPLC, but when we attempted a area under the peak vs. dilution test we got a catastrophic loss of linearity. Values of area decline steadily from the expected value with increased dilution, e.g. at a 1:6 dilution of the standard (1 volume standard + 5 volumes diluent), instead of the expected 0.1667 of the area of pure standard we get 0.079, and the difference gets much worse for higher dilutions.

We are also observing a systematic difference between replicated injections of the same vial, with the second injection producing 20 to 30% less area than the first one. This behaviour was observed in sextuplicate vials, in each one the second injection was lower. The difference was statistically very significant in a t-test and was observed at two different concentration levels (1:15 and 1:72 of the pure standard).

I have no clue at what is going on, please help me!

Thank you in advance.

[unmgvar]

you column is probably absorbing the virus to it,
it is a common problem in SEC columns,
one solution is to overload the column with the standard and hope it helps until you wash the column.

another solution is to look for another phase, maybe SEPAX, they have good columns and they have also what they call a C sec phase
another solution for you is to try and see if another buffer or composition with small amounts of organics help you get rid of this sticking effect,

one final solution for you if you find no way to solve this, is to now that this is the actual mistake occuring and so use 2 linear curves, one for the low % and one for the assay,it is the main rule in small molecule pharma.

for the second problem, how much are you injecting and in what type of vials? do you inject only twice or more. as a first crazy guess, you are using a conical vial and there is not enough solution in the vial or the needle is set high and cannot pick all that is present in the vial,
but more details would help

[mspittel]

> you column is probably absorbing the virus to it,
> it is a common problem in SEC columns,
> one solution is to overload the column with the standard and hope it helps until you wash the column

Possible but unlikely. The concentrated standard is run several times for repetability before running the linearity curve with diluted samples. If there were binding sites in the column, they should be saturated by the time the diluted samples are injected.

> another solution is to look for another phase, maybe SEPAX,
> they have good columns and they have also what they call a C sec phase
> another solution for you is to try and see if another buffer or composition with small amounts of organics > help you get rid of this sticking effect,

We'll investigate if some quantity of organic helps, but the virus is very labile so it would not be very high. the TSKgel columns had been used succesfully with other picornaviruses

> one final solution for you if you find no way to solve this, is to now that this
> is the actual mistake occuring and so use 2 linear curves, one for
> the low % and one for the assay,it is the main rule in small molecule pharma.

We are indeed using THREE curves: one for the "low" 5 µg/mL level, ranging from 2 to 10 µg/mL, other for "medium" 10 to 50 µg/mL range and a third for the "high" range from 100 to 500 µg/mL. The problem is complex because there ar not certified standards so we had to purify our own to homogeneity, characterize its purity by biochemical methods, size by DLS and electron microscopy and once it is showed to be homogeneous and produce a single peak we used a double beam spectrophotometer to assess its concentration which come in close agreement with the value derived from the area for the pure, concentrated standard. We are not going to use a calibration curve in normal operation but derive the virus concentration from its published absorptivity coefficient at 254 nm, so it's critical that no virus is being retained in the column.

> for the second problem, how much are you injecting and in what type of vials?
> do you inject only twice or more. as a first crazy guess, you are using a conical
> vial and there is not enough solution in the vial or the needle is set high and
> cannot pick all that is present in the vial,
> but more details would help

We use standard flat-bottom clear glass vials and we make sure that the needle deep and residual volume after injection ar plentifull. We even measured the loaded and remaining volumes in the vials to make sure all the volume is accounted for. The problem is observable with both 100 µL and 500 µL injection volumes.

Thank you for the feeback.

[unmgvar]

ok the stuff is complex and requires further reading,
but the last part should be easy

for the injection volume, especially if you are using small amounts in the vial,
the height of the solution of the vial has a force, based on MGH,
so each time you inject 100ul or more the solution level drops a lot and that force is less, so the speed of movement of the sample being sucked in the HPLC is less.
2 things can help you:
1. decrease the rate at which the syringe draws the sample. cut it by half at least.
2. in many HPLCs there is the possibility to delay the needle in to stay longer in the vial, to allow for the sample solution to keep moving inside the HPLC. this parameter is generally called, needle delay time or something close to it. change it to double the time.

[sepscientologist]

If you are concerned that the hardware is not working correctly you might try eliminating some of the
chemistry issues. Try injecting anything something relatively polar in water solution at a couple of dilutions.
Maybe 1% acetone and 0.1% acetone in water. Should get big peak at void and 10% smaller peak at void.

[mspittel]

> for the injection volume, especially if you are using small amounts in the vial,
> the height of the solution of the vial has a force, based on MGH,
> so each time you inject 100ul or more the solution level drops a lot and that
> force is less, so the speed of movement of the sample being
> sucked in the HPLC is less.

I'm not quite sure I got that. By MGH you mean the hydrostatic pressure difference between injections due to volume change?

> 2 things can help you:
> 1. decrease the rate at which the syringe draws the sample. cut it by half at least.
> 2. in many HPLCs there is the possibility to delay the needle in to stay longer in the vial, to allow for the sample solution to keep moving inside the HPLC. this parameter is generally called, needle delay time or something close to it. change it to double the time.

Didn't get that either. The mettering device sucks the proper volume into a loading loop and the sample doesn't "keep moving inside the HPLC" until the needle withdraws and close on the needle seat.

Thanks a lot.

[unmgvar]

try in the sampler setting
change the syringe speed.
simply make it slower
high injection volumes have a bad repeatability because samplers by default are set to inject for 5-50 ul
the result is what you describe, a lower area from injection to injection
the syringe gets the right amount but the sample simply does not have enough time to move up the needle, especially viscous or "heavy" samples

[unmgvar]

what was the conversion factor used to transfer the application from the GE column to the TSK column?
what are the dimensions of the 2 columns?