Dear John,

Thank you for your detailed reply.  In my earnestness to respond to your
previous suggestions, and my desire to find a suitable blocking reagent
for NBT, I forgot to mention that we have been up and running 4-5 robots
per day for over a year, finishing over 10,000 genes in the mouse genome
with a great number of positive (and negative) control probes that
correspond exactly to the published in situ hybridization and microarray
literature!  We will be coming out with publications mining these data,
and will have a presence at the Society for Neuroscience in November.

Point number 3 is the most important.  Could you recommend someone to us
that could help with the NBT analogue idea?  Maybe a biochemist or
chemist?

Number 4 also has a question for you.

My responses to your 11 thoughts:

1.  Yes, we had initially, and are having, success with the method you
mentioned.

2.  No, all reagent concentrations were initially chosen on the basis of
RNA in situ hybridization experiments (tyramide-avidin-biotin-BCIP/NBT)
using 25 um frozen sections with RNA probes designed according to
published literature, and then compared with radio-labeled probes, which
showed concurrence.  Yes, the excessive monoformazan deposits are
probably just a result of using concentrations of reagents and reaction
times that were initially optimized for detection of the lowest levels
of target mRNAs.  We cannot monitor the colorimetric reaction for every
probe, so we established an incubation period for all slides and all
probes of 40 min in the BCIP/NBT/Levamisole buffer(Tris-HCl, NaCl, and
MgCl2, Tween 20), which involves 2 x 15 min incubations and 1 x 10 min
incubation, each incubation being a new aliquot of reagents.  Certainly
this is excessive for many probes, and leads to background issues
including monoformazan, but it is standardized for all probes.  The
literature states that only in the ideal reaction does 2 moles of
indoxyl fully reduce 1 mole of ditetetrazolium to produce 1 mole of
insoluble diformazan.  More often than not, the result is 2 moles of
half-formazan (Smejkal and Kaul, JHC, 2001; original reference not
available).

3.  Yes, I am glad we agree on how NBT could bind to white matter
preferentially since is it a lipophilic cation.  My original question
was how to block this binding.  We already use several blocking steps
(please see the web address I gave you) designed to inhibit non-specific
binding of protein reagents, but no steps designed for the blocking of
NBT to the tissue.  Do you know of any analogues of NBT that are not
substrates or inhibitors of alkaline phosphatase, are colorless, and
would not be reduced by BCIP or a general reducing agent?

4.  Yes to A and B.  For C, we tried to wash excess or non-specifically
bound substrates and products after the in situ hybridization (ISH) is
finished, but water, acetone, or 95% ethanol without success (no
decrease in ISH signal).  We will try acid-alcohol washes soon.  Just
after inhibiting and washing off the completed colorimetric reaction, we
treat the sections with 4% paraformaldehyde for 10 min.  Do you think
this step could serve to fix the non-specifically bound NBT and
associated products to the tissue so washing them off after ISH becomes
even more difficult?  What kind of washes would you recommend?

5.  Yes, please see my comment #2.

6.  Agreed.

7.  Okay.

8.  Incorrect; see my reply #2.

9.  Miscommunication/misunderstanding; see my reply #2.

10. Miscommunication/misunderstanding; see my reply #2.

11. Miscommunication/misunderstanding; see my reply #2.

Thanks!

Ralph

-----Original Message-----
From: John A. Kiernan [mailto:jkiernan@uwo.ca]
Sent: Tuesday, October 18, 2005 8:28 AM
To: Ralph Puchalski
Cc: histonet@lists.utsouthwestern.edu; Johnson, Teri; Gayle Callis
Subject: Re: [Histonet] Pink precipitate is monoformazan from BCIP/NBT
reactiononISH?

Dear Ralph,

Eleven thoughts follow.

1. You state that "The incubation periods and concentrations
were optimized for detection of the lowest levels of target
RNA expression." Does this mean you initially had success
with in situ hybridization, biotinylated tyramide
amplification
of the peroxidase label and further amplification by an
avidin-alkaline phosphatase step and BCIP-nitroBT?

2. Were the riboprobe concentrations, times, temperatures
etc chosen on the basis of some other method? If they were
based on stained blots, especially if the hybridization
signals were detected by a different method, the same
conditions probably would not be optimal for sections of
tissue. In a blot, the mRNA concentrations are very low
because each band has an area of several square millimetres.
In a section the local concentration in a cell that's busy
expressing a gene is going to be much higher, with the
expressing cell surrounded by other stuff (mostly parts
of other cells
, in brain tissue) that do not contain that
mRNA. [Sorry for that very long sentence!]
    Your sections might be receiving too much riboprobe
(nonspecific attachment and staining everywhere
, albeit in
the wrong colour. Alternatively the sections might be
failing to bind any riboprobe, and the pink monoformazan is
just a consequence of excessive incubation in the last stage
of a ridiculously complicated method that has detected
nothing.

3. NitroBT is a large lipophilic cation (Horobin 2002;
Chapter 5, p.166 in the 10th edition of Conn's Biological
Stains). As such, it can be expected to adhere more to
white matter than to grey matter in either frozen or
paraffin sections

4. When you deliberately reduced the
nitroBT with ascorbate, in known hybridization-negative
sections, you proved that (a) your alkaline phosphatase
detection method could generate blue diformazan,
(b) the natural disposition of the nitroBT was what
you would expect from its well documented chemistry and
histochemistry, and
(c) you may not washing sections well enough to remove
weakly bound reagents.

5. Have you seen localization of any well studied mRNA
in a well documented site in the brain? If not, your
procedure will never localize the sites of transcription
or translation of any genes.

6. You  state "First, a succession of blocking steps
inhibits endogenous protein activity from interfering
with ..."  That's standard practice in immunohistochemistry
and in situ hybridization, but the blocking reactions
need ro be understood by the researcher. Some blocking
methods can, if misused, provide wrong false-negative
controls.

7. I haven't yet visited
http://www.brain-map.org/pdf/ABADataProductionProcesses.pdf;jsessionid=9
FF15AEC860CC06CC8ECD7E42FF46402
to review the technical details of the methods carried
out by your robots. I may have a look later in the week, but
don't expect a detailed review by way of Histonet.

8. My guess is that your in situ hybridization
technique, despite its multiple amplifications, is
detecting nothing when applied to sections.

9. A possible explanation for the pink nothingness
may be your "succession of blocking steps." Is there
one that extracts mRNA or interferes with subsequent
hybridization? Protein blocking procedures might
prevent the oxidation products of biotinylated
tyramine from binding to the tyrosine side-chains
assumed to occur universally in animal tissues,

10. Before setting the robots to work on 20,000 genes,
try to get the method to work under human control for
5 genes. If difficulties arise with even one of the
five, there can be no justification for investing
4000 times as much money in applying the technology
on a larger scale.

11. In every kind of histological or histochemical
staining the chance of a fatal error increases with
the number of steps in the technique and with the
number of slides being processed. You need to get
your method right for individual slides and then
groups of 10, long before attempting to process
800 slides in a day.

John Kiernan
Anatomy, UWO
London, Canada

---------------------------
Ralph Puchalski wrote:
>
>  Dear John,
>
> Thank you for your advice.  Yes, we use digoxigenin (DIG)-labeled RNA
> probes (riboprobes) in the in situ hybridization of frozen sections to
> map gene expression throughout the mouse brain.  Detection of bound
> riboprobe is a multi-step procedure.  First, a succession of blocking
> steps inhibits endogenous protein activity from interfering with the
> colorimetric enzymatic reactions.  The colorimetric reaction itself is
a
> four part process, starting with the addition of a
peroxidase-conjugated
> antibody directed at the DIG-UTP hapten incorporated in the bound
> riboprobe.  A Tyramide Signal Amplification step is utilized to
maximize
> sensitivity.  In brief, biotin-coupled tyramide is added to the
tissue,
> resulting in the formation of multiple activated tyramide molecules
> through the activity of each bound peroxidase-coupled antibody
molecule.
> These highly reactive tyramide radicals bind to protein residues in
the
> vicinity of the bound riboprobe, thereby resulting in an amplification
> of bound biotin molecules available for detection by up to a hundred
> fold (relative to the number of bound antibody molecules).  These
biotin
> molecules are then bound to NeutrAvidin-AP, the third step of the
> colorimetric reaction.  Alkaline phosphatase (AP) conjugated to the
> NeutrAvidin (NA-AP) enzymatically cleaves the phosphate from
> 5-bromo-4-chloro-3-indolyl phosphate (BCIP), and two of the resulting
> indoles undergo a redox reaction with nitroblue tetrazolium (NBT) to
> produce a blue particulate precipitate at the sites of riboprobe
> binding.  We process about 800 slides (1600 sections per day), and you
> can visit our site for more details of our method and product at:
>
http://www.brain-map.org/pdf/ABADataProductionProcesses.pdf;jsessionid=9
> FF15AEC860CC06CC8ECD7E42FF46402
>
> Processing 800 slides per day in order to cover the entire genome of
> about 20,000 genes necessitates the use of robots.  The slides are
> incubated in only 300 ul of a particular reagent solution in chambers
> mounted on an angle.  The incubation periods and concentrations were
> optimized for detection of the lowest levels of target RNA expression.
> The conditions were established for the entire project and major
changes
> cannot be made.  The system was set up so that all sections receive
the
> same amount of all reagents, including BCIP, NBT, and NA-AP.
>
> You commented on nothing dehydrogenases.  Our experiments demonstrate
> that removing NA-AP from the reaction, while keeping all other steps
as
> they are in the control, produces sections that are transparent, i.e.
> without blue or purple staining under bright-field microscopy.  If
> anything, the sections have a very faint yellow tint, the same as what
> we observe in the bottles just after we prepare the BCIP/NBT substrate
> in buffer with Levamisole.  However, if we incubate these slides in a
> reducing agent such as 50 mM alkaline ascorbate for 5 minutes in pH
9.5
> in Tris-Magnesium-Saline, the slides paired to the ones that are
> transparent become stained deep purple-blue (no NA-AP).  Therefore, if
> we are interpreting the data correctly, nothing dehydrogenases
> contribute nothing that is detectable, and ascorbate reduces NBT to
> diformazan.  Do you agree?
>
> The troubling point is that the amount of staining visualized by
> reduction with ascorbate matches the staining by the pink precipitate
> (formed in the presence of NA-AP in our production runs and controls),
> which presumably is monoformazan as determined spectrophotometrically.
> The staining levels are highest in the fiber tracts, brainstem, and
> hypothalamus, but are visible throughout the sections to various
> extents.  The reason why this is troubling is that the staining does
not
> require NA-AP.  The NBT (we will test to be sure BCIP is not required)
> just sticks to these areas, and later ends up on production slides
> (NA-AP+) as monoformazan or pink precipitate, which is our hypothesis.
> Do you agree with this assessment?
>
> How would you go about decreasing the staining of the fiber tracts,
> brainstem, and hypothalamus, and the rest of the section with NBT?  We
> plan to titrate the NBT, but cannot reduce signal levels.  Our
fixation
> and dehydration process does not remove all the fats in the tissue, so
> it is quite possible that the hydrophobic NBT gets partitioned into
the
> lipid phase.  Would you attempt to change the BCIP/NBT/Levamisole
buffer
> (Tris-HCl, NaCl, and MgCl2, Tween 20) used in the incubation with
NA-AP
> for color development?
>
> Do you know if anyone has tried to block the endogenous sites to which
> NBT adheres?  Without knowing how the binding occurs, it is difficult
to
> suggest options.  How would you go about characterizing the binding of
> NBT to the sections?  I wish it were as easy as finding a convenient
> analogue of NBT that is colorless, does not inhibit AP activity, is
not
> reduced, and binds to the same sites as NBT does, thereby blocking NBT
> from binding to these sites.  Any suggestions?
>
> Your advice is appreciated.
>
> Ralph Puchalski
>
> -----Original Message-----
> From: John Kiernan [mailto:jkiernan@uwo.ca]
> Sent: Monday, September 26, 2005 9:42 PM
> To: Ralph Puchalski
> Cc: histonet@lists.utsouthwestern.edu
> Subject: Re: [Histonet] Pink precipitate is monoformazan from BCIP/NBT
> reaction onISH?
>
> Dear Ralph,
>
> Your email is full of abbreviations and jargon.
> Am I right in thinking it's a question about
> localizing alkaline phosphatase activity by an
> indoxyl-tetrazolium method? If so, please provide
> a reference to the original publication and let us
> all know if you followed it exactly or made any
> changes. Part of your email suggests that the
> alkaline phosphatase activity is not endogenous
> but part of an amplification system used in
> in situ hybridization. The pH optima of endogenous
> and label alkaline phosphatases differ.
>
> The later paragraphs of your message indicate that
> you may not fully understand the significance of
> the mono- and diformazan products of reduction of
> nitro-BT. Both colours result from reduction of
> the tetrazolium salt, but you need controls to
> prove that the reduction was by bromochloroindoxyl
> and not by other reducing agents (such as diaphorases)
> in the tissue. Non-enzymatic reduction of tetrazolium
> salts by -SH has been a well known artifact {"nothing
> dehydrogenase") for 40-45 years.
>
> John A. Kiernan
> Anatomy Dept, UWO
> London, Canada.
> ___________________________________________________
> Ralph Puchalski wrote:
> >
> > I am trying to figure out the origin of the pink precipitate in the
> > image I posted on http://www.histonet.org/site_images_frame.asp
> >
> > Please go to the image entitled: Pink Precipitate ver2.jpg.  It is
at
> > the top of the list on 9/26/05, posted by Ralph Puchalski.  To see
the
> > pink precipitate artifact, please open the image and set the scroll
> bars
> > on the bottom and right side of the image at their 1/2 way points.
It
> > is ugly!
> >
> > I think this precipitate is the aggregation of the monoformazan
> > intermediate generated from NBT (after dephosphorylation of BCIP by
> > alkaline phosphatase) that is not completely reduced to diformazan,
> the
> > insoluble dark blue or black precipitate that labels cells
expressing
> > target mRNAs in our in situ hybridization reactions.
> >
> > We don't know how the monoformazan adheres to the sections of
tissue.
> > It appears to be non-covalent due to the tendency of the
monoformazan
> to
> > migrate under the coverslip in the aqueous mounting medium that has
> yet
> > to dry, and form clumps or aggregates or pools as seen in the
picture.
> > The monoformazan is soluble in ethanol so doesn't form pools or
> > aggregates.  But as soon as it is exposed to an aqueous medium, it
> > precipitates.
> >
> > If we mount the post-ISH tissue sections with an organic based
> mounting
> > medium like UV-CureMount (Instrumedics), the monoformazan might
cause
> > the entire section to turn to a brown tint as seen on
> > http://www.histonet.org/site_images_frame.asp
> >
> > Please go to the image Brown Tint 1.jpg.  It is 4th image from the
top
> > on 9/26, posted by Ralph Puchalski.   The lower image is mounted
with
> UV
> > CureMount, and the upper was with aqueous Hydro Matrix.  There is no
> > pooling or precipitation of monoformazan with UV CureMount, but I
> think
> > it does cause the entire section to turn brown.
> >
> > Question:  How do we eliminate this problem, which I believe is
> > monoformazan?  If we reduce it fully using ascorbate, the section
> (later
> > mounted with HyrdoMatrix) turns dark blue or purple, the same color
as
> > our ISH signal.  We have tried washing off the monoformazan with
100%
> > ethanol prior to coverslipping, but only small amounts are removed.
> > 100% acetone also does not work effectively.
> >
> > Please let me know if you have any ideas that might help us
eliminate
> > this problem.
> >
> > Thank you,
> >
> > Ralph
> >
> > Ralph Puchalski, Ph.D.
> > Manager, Process Engineering and Automation
> > Allen Institute for Brain Science
> > 551 N. 34th Street, Suite 200
> > Seattle, WA  98103
> >
> > ralphpu@alleninstitute.org
> > Tel: 206-548-7041  Fax: 206-548-7071
> > www.brainatlas.org
> >
> > _______________________________________________
> > Histonet mailing list
> > Histonet@lists.utsouthwestern.edu
> > http://lists.utsouthwestern.edu/mailman/listinfo/histonet


_______________________________________________
Histonet mailing list
Histonet@lists.utsouthwestern.edu
http://lists.utsouthwestern.edu/mailman/listinfo/histonet