CD8 enigma... Will it take a village? (Part 1)

From: Mario Roederer (roederer@drmr.com)
Date: Wed Aug 28 2002 - 09:35:52 EST


PLEASE! Do NOT reply to the List regarding this email! Reply to me directly.

We have been struggling with a staining "artifact" for a while now,
and I have run out of good ideas of how to proceed.  I'd like to put
this problem to the community, to see if we can solve it together.

If you're not interested in a puzzle (which will be fairly lengthy),
nor in immunophenotyping, then please skip this email.  This is not a
weighty problem, but it is perplexing and my hope is that the answer
may ultimately teach us something interesting.  Not only that, but it
may also be important for laboratories who use FITC CD8 in their
experiments!

What I propose is to lay out the problem as we discovered it, and
then summarize some of the ways in which we have tried to solve it.
Then I would like to solicit suggestions for what is going on and how
to answer the question...  I will collect suggestions for a while,
and, every so often, take the suggestions to the lab test the various
ideas that come in.  Think of it as doing experiments by email.  Your
brain, our hands.  Your nutty ideas, our reagents.

By the way, I would be grateful if anyone out there would reproduce
our artifact.  At least it would let me think that I haven't lost all
my marbles.

I will post the results of intermediate experiments (and/or the final
answer), giving full credit to those who have made suggestions that
were tested.

Of course, it's also possible that I have missed something completely
simple, and I will be inundated with emails pointing out an obvious
solution... in which case this enigma will not be very interesting
and will die a quiet death (as would my ego). Or maybe someone
already knows the answer.  Maybe it's even been published.

But on the off-chance that there is an interesting story to this, I
thought it might be fun to include the larger cytometry community in
trying to track down an answer that has eluded us for several
experiments.

PLEASE! Do NOT reply to the List with your suggestions or comments to
this problem!  Let's not clog up this wonderful email listserv.
Rather, I will summarize suggestions etc. and post them later in a
single email.  (Particularly since many people are likely to come up
with similar suggestions.)  Also, I want to encourage everyone who
has an idea to send it to me--no matter how mundane you think it is.
We've already tried some things that I would have never considered
doing, simply out of frustration!

OK... here's the problem.  (If the figure below doesn't come out on
your email, you can view most of this information at
<http://www.drmr.com/CD8Enigma/CD8Story.htm>.)

Our common compensation controls for years have been PBMC stained
with anti-CD8 conjugated to every color that we use.  Over the years,
we have noticed at various times that the percentage of CD8 T cells
that we get for any given PBMC sample varied according to the
conjugate.  Recently, we decided to explore this issue when we found
a sample for which this was particularly bad.

Specifically, we noticed that the fraction of CD8 cells revealed by
FITC- or Alexa594-conjugated CD8 was much lower than that when
stained with APC- or PE-conjugated of CD8.

We did 2 or 3 experiments trying to understand where the cells went
before checking the obviousŠ in the FITC or Alexa-stained samples, a
significant fraction of the CD8+ cells had "migrated" into the
high-side scatter region, moving out of the lymphocyte gate.  We
hadn't lost the cells, we were simply gating them out.  (See Figure
below).

The next experiment (co-staining with CD14) showed that these cells
were not simply aggregating randomly or even with each other, but
rather specifically with myeloid populations.  (See Figure below).

We have now done a few more experiments testing various conditions.

An early experiment (before we knew about the "migration" out of the
lymphocyte gate) tested the hypothesis that the cells were
aggregating and were either lost during the centrifugation step,
didn't enter the FACS tubing, or were lysed.  So, we stained and
immediately ran on the cytometery (no washing).  We found there was
little loss of cells.  After we discovered the migration effect, we
confirmed this finding by showing that only upon centrifugation did
the cells appear in the monocyte gate.

If you can do an experiment to confirm this, try the following.
Stain different donor PBMC with FITC CD8 or PE CD8 (a two tube
experiment!).  Do you see CD8+ events in the high-side scatter domain
with FITC CD8 but not PE CD8?

The next experiment tested whether or not these cells were randomly
aggregating.  We costained with myeloid markers, and found that the
migrated CD8+ cells were "apparently" CD14+; indicating an
association with monocytes that survives the fluidics of the
cytometer.  It also indicated that this was no random association; it
was specific with myeloid cells.

We tested whether the association is dependent on temperatur was e of
staining (ice vs. room temperature) or the presence of sodium azide.
It is not.  It does not appear to happen with antibodies to CD4 or
CD19; however, these are not as bright so it is not definitive.  CD4
is also expressed on monocytes, complicating this analysis.  In any
case, the association clearly did not occur using an antibody to
CD8beta (conjugated to Alexa 594).

Things we know about the "association" process (the CD8 cells
co-mingling with large side-scattering cells).

(1) It happens with unconjugated CD8, FITC CD8, Cascade Blue CD8,
Alexa 594 CD8.  It does not happen with PE CD8, Cy5PE CD8, APC CD8,
Cy7APC CD8.  (Note that it happens with small-molecule conjugates,
and not with large-protein conjugates).

(2) The effect is "dominant":  i.e., costaining with subsaturating
amounts of Alexa 594 CD8 and APC CD8 causes the association.  (Even
small amounts of Alexa 594 CD8 can cause it to happen, in a
dose-dependent fashion.)

(3) It happens with at least 3 clones of CD8 (OKT8, plus some
commercial clones).  It happens with commercial conjugates from
multiple vendors and our own home-made conjugates.

(4) It does not happen with antibodies to CD8beta nor with antibodies
to CD4 or CD19.

(5) It happens equally at room temperature and on ice and in the
presence of azide.

(6) It is minimal until cells are washed (by centrifugation); it does
not appear to get much worse with multiple centrifugations.

(7) Including excess unconjugated CD8 during the wash does not prevent it.

(8) Most (but not all) of the high-side scatter CD8+ events are CD14+.

(9) The extent of the effect is donor-dependent.  In some donors, it
is a relatively small fraction; in others, it can be a huge fraction
of the CD8 T cells.  In the figure below, approximately 30% of the
CD8's "migrate."

(10) Note that it still occurs to a small extent with APC-conjugated
CD8.  However, it is considerably less than with the small-molecule
conjugates.

(11) The effect is not seen with CD8-dull (CD8a/a) T cells, only with
CD8-bright (CD8a/b) T cells.  (Note, this is observation is
impossible to discern from the graphs below--but it is in fact true).


Figure:  A single sample of PBMC was stained with FITC CD14 and
either with Alexa594 CD8 (left panels) or APC CD8 (right oabeks).
Top panels, ungated.  Bottom panels, CD8+ gated.



This archive was generated by hypermail 2b29 : Sun Jan 05 2003 - 19:26:20 EST