What if.

The speaker.

Would.

You don't.

Like.

I mean you were.

Just what you were four months ago or

months ago.

But.

I thank you for inviting me.

But I'm not sure the mikes on but
maybe that's just for recording OK.

So.

I had trouble finding this room because
I'm so new I don't know the difference

between Pettit and teach so
I was going to the other place.

So I apologize and so
the work I'll be talking about is largely

how we can think about using
nanotechnology for medical diagnostics and

this might be slightly different than type
of talks that you might be accustomed to

and the vast majority the data will be
showing you actually is data collected and

the research I've been conducting
our haggling cutting thing at MIT

in my research lab right now is actually
housing markets Nano and we're kind

of moving us like different direction but
I won't tell you that at least for

the context of this talk the activities
that's going on in that space but

I'd be more than happy to let you
know what we're thinking about now

if you drop by our office or
lab and I'll show you around OK.

So like I said we're very interesting and
think about how we can use nanotechnology

for medicine applications now one of the
first areas that we've been heavily and

vested over the last five four five years
is in the area of cancer diagnostics and

thinking about how we can develop
better tools to detect cancer is for

example at early stage when treatments we
know are most effective and the question

of course is that we want to develop these
technologies to help clinicians may.

Clinical decisions that are important
to the outcomes of patients now if you

look at a typical just a cartoon
of disease progression in for

example a typical patient with cancer
he or she might be healthy for

several decades before they develop
a two or they would go in for

some therapeutic intervention they
were entering into remission stage and

eventually some of the subsets of
these patients will indeed recur

Now specifically what I want to talk to
you about today is the use of biomarkers

in medicine now biomarkers are very simply
a molecular indicator of disease and

I'll tell you more about the different
flavors of biomarkers are currently use

but the most famous ones right now
are ones that are used to diagnose or

to stratify patients at the time in which
the disease is actually detected OK

because it turns out cancer
is very heterogeneous

there are many different flavors of cancer
if we say lung cancer there are many

actually different subsets genetic
subsets of lung cancer that actually

respond very differently to different
treatments so here I'm showing you two or

three different examples of the type
of markers that we're looking at

in order to help personalized
medicine one of these for

example is there her two new amplification
this occurs in breast cancer patients

if you have this particular subset you
are most responsive to E.G.F.R. or

excuse me or septum based therapies for
example we have examples like.

You have the V three E.G.F.R. mutation
glioblastoma so that we are definitely

identifying new and emergent markers
that helping us to stratify patients at

the time of diagnosis now if you zoom back
even further and you asked well what other

flavors of biomarkers are there that is
currently used in the clinic the earliest

ones who are very simple snip
are a Terry hereditary mutations

that you can look at the most famous ones
are the BRACA one bracket two snips and

these are of course if you're a female and
you harbor these you actually should.

Go for a more intensive screening for
breast cancers as well as on.

The cancers because you
are actually at a two or

three higher risk over
the lifetime of your.

Over your lifetime of
develop new types of cancers.

Now also as we're learning we're actually
surprisingly learning month best to use

different types of biomarkers also not
a good example of this is the prostate

specific antigen test the P.S.A.
tests I think all men over forty or

forty five be very you know
this is common vernacular

this is something that you monitor it's
a blood test now it turns out that P.S.A.

has a very poor negative predictive value
and what that means is that you could have

a high positive test but it doesn't really
tell you whether we should take action

because there are or indolent types of
prostate cancer when she can just wait

really won't kill you during your lifetime
and there are more aggressive types of

prostate cancers which require immediate
intervention and so it turns out that

United States task force agency recently
has no longer recommended P.S.A.

as a screening test because of this
poor negative predictive value but

it's still not a bad biomarker because in
the context of recurrence monitoring and

high risk patients that already have had
prostate cancer it turns out to be a very

sensitive biomarker for that situation
now the question of course is

that within this space we're still missing
a lot of biomarkers for early detection

and a fancy word pharmacodynamic
biomarkers twitchy sensually means markers

that became used to monitor achievement
efficacy of drugs in real time OK So

the question of course is why is this so
challenging for

us to uncover new biomarkers if you
compare the rates in which the F.D.A.

approved the rates in which new drugs and

therapies are approved drugs vastly
outpace that diagnostics and yet

diagnostics if you just sit back and think
about it you might intuitively think or

guess that it might be each actually
easier just to develop a very simple test

to look at that differ.

Type of diseases now the challenges
are several fold and

here I'll just go through three what I
think by the major bottlenecks in this

area the first is if you think of
the different types of data are currently

being targeted as tests OK there are many
different flavors of molecules that we can

look at we can look at metabolites more
classically we can look at proteins but

of course there are even new and
more exciting biomarkers even such that we

can look at for example single individual
circling tumor cells as a way of them

forming a disease state now the challenge
I would pose to you as an engineer is that

there is a tremendous diversity in the
properties of these molecules OK And so

as an engineer every time are interested
in probably something else we have to

develop an entirely new and a local
platform to fish these guys out so that's

pretty challenging the second aspect I
posed to you is that as biologists as

these particular maybe you know trained in
biology maybe not full fledged biologists

is that the limitations of this is that
we're essentially limited by what nature

provides us OK So if a disease doesn't
happen to produce a unique protein

we have to look at some other subset of
molecules in order to do a diagnosis and

so that makes it challenging so together I
like to collectively clump these together

and say that these are and markers
because they are naturally occurring in

healthy and disease systems now the second
aspect the second bottleneck is that most

of these tests most F.D.A. approved tests
are single marker tests OK we know that

single markers are prone to high rates
of false positive false negatives OK And

then too late the way that you can
think about this is that the more

you know we have first names we have last
names we have middle names to use or for

my name is Gabriel manual name you know
embarrassing is actually Abner So if.

That's pretty darn unique there's
probably only want to be in the whole of

the United States but if I say if I
just use Gabe there's a lot more people.

Nice States that's named gave Gabriel also
OK so what that means is that the more

parameters of the more markers that
you are able to modify the same time

the more specific test will be for
a particular disease

now this is used all the time in the good
example this is in prenatal screening

we have a three biomarkers serum tests
which is a has a very good specificity for

a few employees and this would be like
Down syndrome OK And then if this kind of

specificity is only attained by looking
at multiple markers at the same time.

OK And the third element I want to
very briefly share with you is that

most of these tests are designed
to be blood tests and

this is great because blood obviously
babies all their major organs and

so if it's a diseased tissue is
spitting out these unique molecules or

any cares in the blood you can just
take the blood sample and analyze it and

hopefully be able to.

Define what that disease is but

the challenge of blood is that
it's a very complex medium and

if you think about it especially you want
to do early stage diagnostics where tumors

are on the order of say five millimeters
what happens is that these tumors as

they are shedding these bomb markers in
the blood what happens it goes from a pea

sized volume into five liters of blood so
the first thing that happens is you

have a biomarker that's enormously diluted
into our blood pool so it makes it very

challenging to detect Now of course there
are many bottlenecks to this process and

that's just one we have to consider for
example how quickly tumor cells grow

how quickly they can stash we
make a protein biomarker for

example which is limited by cellular
machinery a cell can only make proteins so

quickly and furthermore as these protein
markers for example shown in green or

circularly in the in the blood they
have a certain half life they're not

stable OK there are clear
by normal mechanisms and

so there are a lot of great constants
here that we have to consider and

what Sam Gambhir as Stanford
did was he did just that he.

Built a mathematical model and
he asked how sensitive can a typical blood

biomarker actually be in carrying cancers
especially early stage cancers and

he took into account what we currently
know about growth rates about

the stability of these bombers and their
results are actually quite striking and

what he's showing you here on the X.
axis is time until detection time with

Genesis where point zero we would
assume that's one tumor cell and

then we're worse we modeling how gross and
what he found it at least for

ovarian cancer is that with our
current technologies using a blood

test it would take anywhere
from eight to ten years for

our standard clinical Eliza
test to indicate disease and

by this time is certainly not early stage
disease these tumors are quite large

they're not wanted to send reason diameter
and it really does not eat fancy or

medical imaging platform such as in
this case transrational ultrasound OK So

this is where we are with all
these blood biomarkers to you so

the bottlenecks that's we're very
interested in addressing and so

our approach is that since there are so
many potential limitations with a natural

biomarker deaths very hard to circumvent
right because these are biologically

constrained we were we asked ourselves
a question well could we make a market

that synthetic OK if it's synthetic
we can engineer a way a way

around some of these constraints that I
just showed you share with you and so

our thoughts was well instead of
an dollars a spinal marker let us make

an exogamous agent nanoparticles and I'll
show you how we've built them later on but

it's an exhaust agents debts would
produce a disease contrasts so

we're going to make our own disease

detection signal instead of relying
on something that's natural

the second aspect I want to share with
you is how we can take a single test and

build a multiplex library OK And again

like I said earlier the reason why I want
to do this is this would increase the.

Our test and hopefully not just be able to
differentiate different types of flavors

of cancers but
allow us to do staging as well and

the third aspect I want to share with you
is how we design synthetic markers so

that they are not detected from the blood
but rather from the urine Now this was

circumvent some of the limitations I
mentioned earlier about biomarkers

dilution in the blood and what's nice
about urine is that it's concentrated and

it's a downstream filtration
product of blood so

ideally you would potentially
hath assess ability of blood but

in a more potent and
concentrated a solution OK So

what are the synthetic markers.

If you can survive this cartoon you
can survive the rest of the OK so

synthetic biomarkers are very simply
nanoparticles and I show here just

one formulation in which these these are
particles and they happen to be clustered

together but you really can't take any
type of nanoparticles scaffold and

use their particle states recreate
on the surface with peptides

these peptides have two unique domains
associated with them showing and

color is a region where it's pretty
sensitive meaning that they can make

leaves by a class of enzymes called
produces now produces If you don't know

basically ends of magic scissors they're
molecular scissors and they go in and.

The cleavage peptide OK so you can
cover this part of these particles with

Cleave more substrates and to the second
half of this domain you can couple of mass

barcode and I'll tell you later
on exactly how we design this and

the function of this mass barcode and the
idea here is that you can pull together

a multiplex solution of these different
flavors and then a particles sense of

the different types of produce this into
one injectable formulation you infuse into

the patients and we use the nanoparticle
carious because their function is to

essentially deliver these peptides to
the disease site here I'm just showing

you it's as the liver metastasis but this
actually this phenomenon that holds true.

For many different types of cancers.

And here within the disease Mark
environment what we're asking

the nanoparticles to do is to present
these peptides to the local market

environment so that these protests
which are specific for disease for

specific for
cancer were common and cleave and

shave these parts off the surface of
the particle now a very interesting thing

happens now is that these peptides that
there are now very small OK previously

there are a couple that small is in
relative terms there are a couple to

nano particles in these particles
happens to be about fifty nanometers so

they go from fifteen and a meters to two
diameter measured in angstroms and so

what ends up happening is that when these
fragments enter into the circulation they

actually very quickly filtered and
concentrated into the urine OK So the idea

here is you can collect a urine sample and
because each of these peptides have been

prematch labeled we can come in with mass
spec quantitative mass spectrometry and

be able to tease out the cleavage the
presence of these cleavage fragments as

who also relative abundance of
these fragments in the urine OK.

So synthetic markers are very simple like
I said they are made of produce sensitive

couple of nano particles the very first
test you can do is to design sequences

of sequences to sense the different
flavors of produce is in the way that you

would do that is to make peptides which
are terminate at the end before force and

what happens is when you pool when you
conjugate these peptides to a nano

core it tends to promote home of
quenching between the floor so

this is a dim fluorescent state you can
incubate this with the purpose of interest

you a clear peptides off and
now the floors are free to fly us and

using this very simple four channel gas
and you can go about screening different

rates looking for relative specific
sequences against Proteus of interest

here we've designed a peptide against them
which is a blood Proteus you can see there

a pretty strong activation
fluorescence specific verse.

Other types of.

Blood as well as the and then T. family.

OK So like I said earlier since
their biomarkers a very simple

they're peptide to nanoparticles and
we were very interested in

figuring out exactly how these individual
components interact within their living

animal to produce a urine signal now this
first piece of data is what would happen

if you took three peptides to label with
a reporter for us or an importer and you

inject it into Disease and Control animals
you would immediately see that the liver

excuse me the bladders The lighting up and
this because the peptides are small and

that just being secreted very rapidly into
the into the bladder now what would happen

if you conjugate floors to the surface
of the nanoparticle is you would see

in this case the first thing the lack
of fluorescents in the bladder and

this is like I said earlier because
these particles are large so

they're they don't actually gets created
they're circulating in the blood and

they eventually get scavenge by resin
that monocytes America faces in

the deliberately left nose and that's why
the liver in this case is lighting up and

the third piece of complimentary
data I want to show you

is what would happen if you injected
particles now particles in which the for

force are labeled again the first
thing you would notice is that

the see the bladder for
the disease animal is lighting up OK but

the control animal is not.

And so what this tells us is that at least
in this very broad overview of these

animals is that there are some produce
in the disease animal that seems to be

cleaving these peptides and
these fragments are kind initiating this

pharmacokinetics which that they have now
after clears are going into their urine

now to give you more direct evidence of
this process actually happened I want to

show you data that we collected from
a mouse model of fibrosis excuse me of

thrombosis not from bosis is the formation
of blood clots and these blood clots

are usually found in intravascular and
this particular mouse model.

You can monitor the disease burden by
infusing fluorescent fiber engine and

just a reminder Firebird engine is
a precursor for fiber and clots OK So

when you have active clot formation
Fairbridge it will be clear and

incorporated into the newly formed clots
and so by using this very simple reporter

you can see in this case in mice that
were induced with on both as a very sharp

regulation induction of these blood clots
within the lungs as quantified here on

the right hand side now you might ask what
would happen if we injected these Flora

genic sensors like I said these
particles are in the current state so

if they are clear they will be activated
in the flush and should go up in this case

if you come back to the lungs in
the thrombotic animal you can see

a significant increase in particle
activation with the lungs that we did not

observe in the control animals Furthermore
as you start looking downstream you look

at the kidneys you can see a fluorescent
we observe increase in fluorescent

activity in the kidneys and furthermore
lastly when we looked at the urine for

the presence of these clippers fragments
we indeed observed a significant increase

in the level of fluorescents in
the urine versus the control animals

Now we also repeated these experiments in
the presence of by a valid route in by

a valid route in is a direct inhibitor
OK so in the presence of a direct

an inhibitor we were able to abrogate
any differences in the urine signal and

so collectively what this
data tells us is that we can.

Get the synthetic markers are indeed
activated at these sites the fragments

are going to the urine and this activation
is really dependent on the activity of.

OK.

Now this platform potentially can be quite
sensitive because there are two major Most

of signal amplification The first is that
we're targeting produce is you know we're

not looking at the abundance of produce as
we're actually relying on produce activity

OK so produces our catalytic
one produce concrete.

Thousands of peptides per unit time so

there's your single application the second
part that potentially could be very

sensitive is again we're not looking in
the blood where these fragments looted but

we're actually looking in the urine Now it
turns out every five to ten minutes all

your blood pool is filtered through the
kidneys OK this is just normal physiology

so what and your kidneys of course is
responsible for you absorb the water so

if there is a point source generating some
signal somewhere in your body releasing

these reporters it's filtered every five
to ten minutes and into the kidney and

the kidneys are to draw in the water so

essentially concentrations reporters
into the urine so now compare and

contrast this platform versus a normal and
dodges biomarker shown in green again

once they're from the tumors they have no
ability to amplify their own presence and

furthermore especially if we're doing
a blood assaye they're diluted in blood so

I just want to compare and contrast to
these two different types of approaches.

And so we came very interested in this in
trying to decipher whether this approach

might be potentially more sensitive
than traditional blood markers and so

we focus on one particular protein it's
called C.D.A. it's currently F.D.A.

approved for colorectal cancer.

And what's striking to us is if you look
at all this stablish available human

colorectal cancer cell and
some A.T.C.C. plotted on the X.

axis and you ask how much protein can
these cells make per unit time we

found that these rates varied by over four
orders of magnitude in difference and

now each of these cell lines is derived
from a different patient and so

this kind of gives you a glimpse into the
heterogeneity of these types of cancers

and it can you can already kind of scratch
your head and say well if I develop

a blood test for CA How am I supposed to
account for this vast diversity OK And

so we decided to do was to do fairly
competitive comparisons so we took a cell

on this one seven forty because it's
produced a clip one hundred for

the book the median value expected in
humans now if you implant two cells.

Into my state form tumors expect to
expend exponential growth rates and

furthermore Now if you ask what
are the levels of the CA protein

these tumors can produce and
that's detectable in the blood and

what we found is that we did not observe
any significant elevations in CA until

day thirteen OK when the tumors
were fairly reasonable

size about three thirty
millimeters cubed in general.

The first thing that we want to assess
was whether our synthetic markers could

indicate the presence of cancer and we did
this very simply with particles you can

see that the bladder is lining up here and
furthermore we want to head and quantified

the elevation in the urine signals
here in the left you can see this is

the scatter plot for the control animals
which did not have tumors in the presence

of tumors there's about a two fold three
fold increase in the urine signal and

furthermore again this
signal could be depleted or

reduced in the presence of a murmur start
in this case is a broad spectrum M.M.P.

inhibitor because in
the presence of tumors.

Plus this inhibitor we see a signal
go down it suggests that the cleavage

activity or the production of
this urine signal is due to

the family of Proteus is typically
implicated in cancer OK And

like I said we wanted to do a head to head
comparison and so in this model we infuse

nanoparticles I ten which remind you
that there is no significant difference

detectable difference in serum in
the concentration of the CA protein.

And yet in our system because of
this ability to amplify signals

we were able to observe a significant
elevation in the urine.

Signals at ten we want further to quantify
the the productivity of this assay and

you can do this using a receiver
operator characteristic curve and

basically the way that this
analysis works is that.

You're looking at the true positive rates
versus the false positive rates as your

varying eight discrimination threshold
OK And so the way that this works

is the perfect classifier the perfect
diagnostic what always give you an area

under the curve of one OK It's never wrong
now a random classifier will give you

an area and perfectly five because there's
no better than coin flip OK And so if we

looked at our serum samples we observed an
eight point six one four C. A which again

like I said this is the F.D.A. approved
biomarker but at the same time point we

were using our synthetic bomber protests
we observe a significant increase in

the area in the curve meaning that we
had developed a more predictive assaye.

For mice to indicate colorectal cancer.

Now like I mentioned earlier do you
start experiments were done with

a cell line we which we chose for
a very competitive

type of comparison now you might ask
what would happen if we chose a cell and

this happens in the clinic in which
the patients tumors are not or

secreted very low levels of CA What would
happen in that situation now it turns out

that if you take this and we chose for
you know because it's making

this year product three fold below the
median value you can do we piece the same

set of experiments you see that in my
state grows exponentially but perhaps

predictive of a significant reduction in
CA production versus tell us once and

for two cells we were not able to observe
any significant elevation in serum levels

of C eight even up to about a month
after we implanted the tumors and

these are very large tumors they're so
large we have to essentially sacrifice

some eyes because it's not really humane
to let these tumors grow that large but

conversely infuser nanoparticle
test that they thirteen

we very easily were able to
discriminate a significant elevation in

the urine samples in
relatively small tumors and so

collectively what this
data tell told us is that.

We potentially because of this ability
to amplify signals looking at produce

activity and look at the urine we
might have a way of circumventing

the traditional limitations associated
with the blood biomarker even having

the ability to indicate cancers that
normally don't produce a typical blood for

us to test OK.

Now the second part I want to share with
you is how we think about multiplexing

a platform that I've just shown you.

Why do we even want to do this now because
our platform is targeting Proteus as it

turns out that there are many flavors of
producers out there aren't you know in

codes over five hundred different types of
purchases there are actually more produces

than kinases in the body half of them are
extracellular which allows our platform

to detect Now if you look at
complex disease such as cancer

this is a very cursory and by no means
exhaustive list of the different Proteus

is that you can find there are implicated
in different stages of cancer and

so because of this complex see if you
just take one probe one sequence and

you you know interrogate a subject most
likely you would not be able to represent

the complex the the underlying biology
the complexity that's inherent within

that disease state and so like I mentioned
earlier what we're very interested in

doing is thinking about ways of doing high
throughput measurements OK how do we think

about doing a multiplex measurement
here in this system and in.

Our design criteria is to think about how
we can incorporate mass bar codes into

these peptides substrates Now the first
thing I want to clarify to you is Why do

we even need a mass barcode OK because
your question might be that if we had

a pretty substrate of interest
that we knew of Shouldn't.

It be sufficient to know where Cleave
because if you knew where cleaved

you can because you are looking
at mass spec simply to

figure out what the molecular
weight of that species is and

then do mass spec and it turns out
that that's the way that the vast.

Majority of in vitro studies are done
you have a putative Cleaver site you

calculate what the medical rate of the
release fragment is and then you do mass

spec Now the problem with this approach
with the evil studies is that very quickly

clear which fragments will be degraded by
produce which comes in and choose down and

so from a fine fragments which you know
where to look you look in the blood or

you look in the urine guess what it's
all degraded in this latter pattern and

most likely single amino
acids are coming out and so

there's it comes confound mass analysis
and so one thought that we were.

We innovated in this space is thinking
about how we can produce mass bar codes

now at the distinction of a mass barcode
is that we use the stereo isomers

OK now if anyone remember from
the standard biochemistry

all of proteins are are handed
all approaches actually coded

OK because there's a sterile center carbon
for each immune acid so it turns out if

you make the right hand that version of
Mean asses guess what can the produce in

the substrate interact with each other
no it cannot clear if the version is so

very simply by switching here represented
by the lower case L to D.M. you know acid

we can create a resistance bar code
if you will on the left hand side.

This barcode sequence really can be
anything but we gravitated towards

this one peptide which actually has
a name it's actually called and

asked the name because
mass spectrometer has.

Found that this particular sequence
I now use is very very well

in a positive ion mass spectrometers and

so why is that you know important for
us well is partially because we're using

mass spec as a downstream analytical
tool the second very coincidental.

Property of the sequence that is very
unique is that it turns out that this type

is derived from fiber engine and it's shed
when fibrinogen is cleaved into the urine.

And so by looking at these studies in the
sixty's and seventy's what we figured out

is that it turns out this
peptide is a very natural.

Creator sequence is not actively
reabsorbed by the body OK so

these two properties made a very
attractive to use as an enviable mass

market and now the second element of this
structure that I want to share with you

is the incorporation.

Funky looking and you know asked and
anyone who is again taking about

should we know that this is not
the canonical any one of the twenty amino

acid that's normally found in the body
This contains a natural fennel group and

if you know natural fennel group
is very sensitive to light

Why do you want to do this because if this
region is clear in vivo and it produces

laddering effect shown here by the by
the cleave lines we don't really care

because now we can take the urine sample
return to radiate by a long way of U.V.

light it will initiate for a cleavage and
it will release our mask for for

analysis and so we have a way of cleaning
up if you will all these messy cleavage

fragments on the other side of the peptide
OK now I'll show it to show you deficiency

of this process you can see the ten M.
peptide mass spectrum on the.

Top here on the upper right after thirty
minutes of U.V. light radiation and

you can see this complete disappear
followed by the appearance of this red

more if you.

OK so now I've shown you how
we go from to how we design

a photo crave a bowl barcode into our
system the question next question will

be how do we go from one to many how
do we build a library OK and to do this

this goes a little bit into into chemistry
sorry apologize if this is a little bit

too much into physical damage but
I find a very interesting.

It turns out that if you talk to
any proteomic person an expert

de you molecules like people.

All the time for mass spec and

I'm showing you one of the structures
here it's called I track molecule and

what's very neat about what these guys
have figured out what these guys and

girls have figured out is that they
have a strategy of incorporating stable

isotopes for so for example if you look at
this is a carbon here you can substitute

a carbon thirteen here OK the substitution
of a stable ice What does it do is shift

by one atomic mass unit but it preserves
the molecule it's a behaves exactly

identical you can do this you can do this
trick and you can the only difference you

can see between these variants of these
four molecules is where they do with

the substitution of the carbon thirteen
and the heavy oxygen eighteen by doing

this trick as you're doing tandem aspect
when you're fragmenting this peptide or

excuse me this report a molecule what ends
up happening is that these four molecules

it turns out if you at the mass of all of
them together without fragments and they

actually have to say more weight OK and
this increases the chance to read because

you can sit there and collect all
the signals you have you know how

to scan a large and the window but
if you fragment this then all of these

variants in the way that these are steps
are disputed becomes very apparent and

you can figure out the relative
abundance of each of these family.

Within each individual within the family
of reporters and so I was inspired by this

approach and I thought well surely be able
to do this same thing for peptides and

it turns out that you can so you can take
peptide sequence I showed you earlier and

split it in half OK and
that's where you do the mass balance and

you can come in here and
it's bluish reported for

example you can consider
this as your natural.

OK here you notice that I substitute
it's a plus one glycine So

this one part they want
to talk master to shift

the next one has plus two plus three
plus four and so on and so forth you

can see that this is an extensible
strategy to build a family of codes.

And so what's interesting is you can pull
all these peptides together ten bar codes

by mass spec by.

The By just mass spec and
they're completely on this thing.

But if you go and do a tandem aspect you
fragment this individual component within

the system would be apparent and I just
want to mention here that these people

are complete quantitative OK so how
much of your reported ten is in solution

is reflective of how high that peak is
OK so it's quantitative aspect and so

here's the fun part you can start thinking
about coupling these bar codes on to clear

of all substrates showing and read OK So
remember these are mass barcode

substrates Now you can couple these to
nano particles and one of the things

you could do is just start monitoring ten
cleavage events at the same time you can

monitor one Proteus to ten vents or
you can monitor ten.

It doesn't really matter but again I like
to emphasize the activity of the top

interactions is really given
by the height of the peak OK.

So we've gone from a single floor Janica
since a biomarker platform I've shown

you how we can use concepts in mass bar
coding to create a multiplex library so

now we have a template set of synthetic
markers So what's an interesting problem

to look at OK so in my previous
training as a post like at MIT

that lab the lab in which we focus our
efforts on cancer now is technology and

the half on liver tissue engineering and
so we had a lot of resident experts and

lab and so
I thought well one of the interesting and

we could kind of merge the two and
so one disease that has a need for

noninvasive biomarkers is in
the area of liver fibrosis and

fibrosis essentially excess
to scarring of the liver and

this could have happened whenever you
have untreated hepatitis B. to see for

example in alcohol abusers and
it turns out that in the in

the western sphere fatty liver disease
because of our fatty diet is actually a.

Markedly rising complex that we will have
to figure out how to deal with in the very

near future now if you look at a typical
histological slice of these levers

characteristics of the fibrosis you can
see an accumulation of scar tissue.

And like I said earlier this process is
currently monitored by invasive biopsy so

you go in there with a needle you
take out a chunk of tissue and

what we really want is a noninvasive
means of marring disease state and

now Furthermore what's very interesting
about fibrosis is that is trim by a very

diverse collection of Proteus is shown
here in the first four panels or

so but also by their natural inhibitors

these tips are the tissue inhibitors
of M.P.'s and so you have and

then piece up regulated but
you also have their inhibitors and so

if you're going to say can we say and
then P two is activity that

is operated it's hard to tell because
it's a balance act between it's

over expression plus the effective than
the hitter and so because of this complex

that we thought this would be a very good
platform actually to do a non-biased

screen to look for biomarkers because of
these competing rates of interactions and

so we implemented a mouse model of liver
fibrosis where you can give mouse give

my diet Chow in which is laced
with a small molecule Did you see

D D D C is a paddle toxin So
after three weeks on this Cho.

These mice will develop liver fibrosis
shown here by serious red standing which

stands to cause and fibers or the scar
tissue within the liver quantified here on

the right hand side you can see that
after three weeks on this on this food

there is a significant elevation and total
amount of scar tissue in the liver and

what's neat about this model is that
now after three weeks if you restore

a healthy normal child to the mice they
have the ability to slowly heal over time

to regenerate into remove
the scar tissue and so here.

Quantified on the right hand side you can
see that the scar tissue peaks around

three to seven weeks but after another
four weeks off of that the child the scar

tissue is being repressed OK So
we have unique windows in which.

There's fibrosis in disease and
resolution.

So within this window we decided to
infuse our templates library of synthetic

biomarkers and these are the type of
data that we can collect by my spec.

One through ten represents different
flavors of peptides substrates that we

infused into this mice and because like
I said earlier we're interested in.

Window and reversal we injected
these particles that zero three

seven eleven weeks now
the first thing that we noticed

is that not all the markers have the same
type of activation kinetics which

I think in itself is very interesting the
second thing is that some biomarkers for

example G one two eight nine and
ten were completely and

informative in the context of fibrosis
state did not deviate from the baseline

values there were some biomarkers that
were significantly elevated only in

the reversal when those again in the seven
eleven when the seven long weeks.

And there was one biomarker in particular
that we found very interesting because it

was significant elevated throwing
the fibrosis window but not reversal and

so it seemed like we had
a template's library we did a.

Blind screen and we seem like we were able
to fish out one peptide that was specific

for fibrosis could differentiate
fibrosis even over.

And so we went further on and we wanted
to cross validated this by a marker cross

validation is important because you don't
want to you know over fit your data for

one model and so we took agent addict
model of the fibrosis this is and

they are to knock out mice you can see
that after about four weeks of birth

they develop naturally develop
scar tissue within the liver and

furthermore when we went into validates

the potential G seven as a biomarker
fibrosis we indeed saw a significant

elevation in urine signal versus a marker
dollars not informative in our day of the.

And so this is a way of cross validating
there are initial hits in our first system

in our thoughts and.

Fibrosis.

One thing that we define very interesting
is if you look at templates of bomb

markers if you looked at the most
predictive biomarker from that pool

the single bomb marker is the best
one gave us an eight of C.

A point seven three so not very good but
now if you ask combination of biomarkers

Like I said earlier multiply X. we were
able to show that when you looked at

a marker combination this could again
significantly increase the area under

the curve create a more
predictive fast say for us.

OK.

In the last ten minutes the show so

I just want to talk a little bit about the
advantages of a urine test Now I mentioned

earlier one advantage is that you're
seeing the concentrate in the urine but

if you think about what we have
is we have a platform which is

essentially imaging free we're
not tied down for the need for

a C.T. and M.R.I. etc We have a vial
of nanoparticles that we can administer

of really anywhere OK at the point of
care but what is not really flexible in

terms of the broad utility is the need for
a mass spectrometer and so we became

actually quite interested in whether we
could design low cost diagnostic step

could be used to analyze urine to
analyze these fragments in the urine and

replace of mass spec so that we can
create a platform if it's low cost and

portable that might be employed in
resource limits environments in the global

health applications and
paper is fascinating material we use it

all the time but not really thinking
about in the context of diagnostics

paper has very natural working properties
it's all true portable is ultra cheap and

very stable you can transport any largely
many different types environments and

it turns out that many different groups
have now investigated the use of

paper to do three dimensional markets
even to culture cells on pay.

And so we got very interested in
a particle or type of paper test

is called a lateral flow assaye on
the left hand side I'm showing you and or

a quick this is a saliva swab asset that
was recently approved by the F.D.A..

And this is exactly the same
type of technology that's found

in the home pregnancy test that you
can go to the local supermarket and

purchase the way that a lot of flasks
a works is that you said to have a sticky

piece of paper that you can print
capture anti-bodies on OK and

then you would apply your your your
sample your urine sample on that Dr pad

and essentially Lao the flu to work
across these capture regions and

if your sample has an interest that will
be localized to capture anti-bodies and

you can come back and you can develop it.

Usually with a gold nanoparticle detection
agent Now the reason why we use gold

nanoparticles is because they're naturally
colored they're pinkish reddish and so

in parts and then gives this very
characteristic red huge that we're all

I think pretty familiar with
in terms of paper test.

To make of particles compatible with
detection on paper we had slightly changed

a bit so remember I just described
to you how we design mass and

code mass bar codes here we want
to switch it up a little bit and

we put in little Legan's these are small
molecules in which antibiotics can

recognise to these particles and
so that when

there is produce activity in these reports
are released they can actually come in and

form a sandwich complex OK where the
antibiotic can capture these fragments and

then we can come in with our detection
agent and read out the presence as we

went ahead and showed that this platform
is indeed orthogonal meaning that if you

design multiple lived in code
reporters they show up only.

When you put into and
a bias against is properly and so for

example here you can see that this
law as well as lighting up this

one is not is because the floor scene
is targeting one of these days here.

And secondly we went ahead and

also thought about how we can make
multiplex paper based tests OK So

mass spec is nice because you can look
at multiple Paramor set a time and so

you want to do the same thing on paper in
the way that you can do that you can use

spatial encoding you can very simply
a place to control line somewhere and

reference these different capture and
roll tip to that control line OK so

by distance you can tell which
reporter is localize where.

And so by doing this we can do a multiple
X. paper test so we went ahead and

want to validate these type these
slave to this particular embodiment of

these synthetic markers back in our animal
models we looked at thrombosis which is

again blood clots and cancer and we first
validated that the paper test could and

couldn't detect activity so we incubated.

Nanoparticles with thrombin for
example for thrombosis and

used our paper strips to indicate
the presence of the cleavage fragments and

then we further went on and to validate
that the quarter report did not change or

alter the way these now particles
are distributed within the mice that they

are equally Cleaves and
localized in the urine as our previous.

Formulations and then furthermore
lastly we want to head and took our

portable paper tests and were able to fish
out significant increases in the presence

of these fragments both in our thrombotic
animals and in our cancer animals.

OK so I'll stop here but
just give a summary.

Of.
Showing us that we can design

since the markers to diagnose M.R.
different types of diseases from

the urine I've shown you how we can do
that for fibrosis thrombosis cancer and

I think you will probably kind of see
if there are different Proteus is.

Distributed in a type of disease most
likely we can tailor this platform

to target that disease secondly that
the idea of using and sematic activity

if we're not looking at abundance of
measurement and sematic activity plus you.

We can potentially create
a very sensitive test.

Lastly our ability to monitor multiple
bond markets simultaneously by mass spec

this provides us a way of doing unbiased

you know screenings in
vivo as well as to come.

Together to improve deceased
discrimination and that's Lastly these

are highly engineer ball platforms we've
shown you how we can make the system

detectable by mass spec by paper or
by allies and we have another paper that's

published looking at how we can use
digital to detect these fragments etc.

OK And I do you want to acknowledge
a very talented group of vigils like

I said earlier most of the studies
were conducted that MIT so the bots.

Where I was supposed to talk for
the last couple of years and particularly.

Ventures in Boston who was one
of the earliest Prasoon so

I developed all of these ideas
within the very beginning

my previous mentor Sankey that my
funding sources that girls welcome for

the career were about scientific
interface and the and.

And I'll stop there thank

you.

And that's a great question that
we try kratom in craton and

turns out so craton in is essentially
a by product of many muscle metabolism so

it's normally present in the blood and
then because a small it leaks out into

the urine so often times it's
used kratom then in the urine and

in the blood to normalize concentration of
urine samples craton then is a baseline

measurement where if the individual
is at is exercising or high activity

it dramatically changes Kratz and levels
we did try using craton and it turned out

it wasn't very good as a normalizer in our
system what we eventually end up doing was

using one of our codes as an internal
control and so it's just like in.

And which is a polysaccharide that you
can go to the doctor's office they infuse

in the blood and
they use that as a call that for.

Filtration of the urine we
took that as inspiration and

took one of our barcode center normalized

concentration.

As a good point there X.
that it's really really just.

Hereditary if you will because it
turns out that you know the saying

there was a deep collaboration
with sale there.

Is chemistry at U.C.S.D. and he was the
one who formulated these nano worms if you

will because they look like worms in terms
of targeting they actually seem to have

increased video because of
the cooperation between neighbors but for

this case because we're just looking
at it really doesn't matter and

in fact one of the areas that we're really
pushing is thinking about how we can use

biologics for example antibodies which
are already used in the clinic but

coupling some of these kleagle peptides
on these biologics because now you have

essentially a Theron Gnostic which can
report on its own advocacy in real time so

that is definitely and
direction that we're pushing but

you're right we don't really need to use a

scalpel.

That.

You.

Know.

Another war.

Well we do we do modify the surface so
it's not just peptides we actually

put a layer of paint on it we in previous
studies we've often mines that pick

a length so the longer the peg chain is
actually in that only prevents this Corona

or delays as Chernoff the happening but it
actually does reduce the Kalak efficiency

of the cutting plate as the Proteus is
makes it harder to penetrate past this

layer so we found this you know Goldilocks
if you will length of twenty K.B. which

seemed to surprise somewhat stability
you know stability to the particle but

still allow the Proteus come in and
cut but I think that is actually a very

interesting space to think about how we
can for the author of mice the surface to

to increase activity and
chill background activity.

I.

Mean your.

Own.

You would treat each in biology
there are no stuff functions.

OK So everything everything in biology
has a plateau OK and that's what you what

we did was that we assume that each of the
biomarkers could be modeled by the shape

OK And so we normalize each of the
intensities to a score from zero to one

OK so for a one zero to one marketing
etc Then we added them up OK and

we use that as our risk or analysis and
so that's basically how we did it but

I'm actually very open to different
ways of interpreting the data and

one thing that I failed to didn't have
time to show you is thinking about these

signatures as you would an R.N.A. and
a transcript or analysis you know because

we have many different you know signals
here just like a marker or a or on a seek

you should be able to do clustering
hierarchal clustering and analysis and

I think that So definitely this is not
the only way of analyzing the intensities.

Well.

You are.

Right.

We have a paper that's about to be
submitted in which we developed a P.K.

model of this whole process so

you can very simply model this as
compartments you have a blood compartment

you have a tumor compartment you have a
urine compartment you infuse the particles

to a certain concentration with a certain
proclivity to be cleaving the blood back

on produces as a certain diffusion rate
into the as a certain half life and

so this model of taking those counts
all these different parameters and

what we are teasing out is what is
the importance of circulation time for

signatures for signatures and it turns
out that blood back on activity is

a significant problem if you will for
this platform where you think about ways

of suppressing blood activity and you know
it does kind of connect with the earlier

your earlier question of worms versus So
it turns out that if you put peptides on

protein based they're actually less
susceptible to cleavage there's evidence

as a company in Boston which
has shown that this is true and

it's probably true that the fact that our
body sees proteins all the time and so

having like a cryptically if
you say it's probably shielded

versus an inorganic particle
which is fairly foreign But

you're right there are many different
parameters were teasing that out and

our model would most likely allow
us to optimize this formulation

better.

Than the.

Yeah we're actually.

That's probably the most exciting
aspect of this work is that it is

highly translatable we have we have
three patterns that were issued over

the last five years on this
technology we are spinning a company.

Although it's being you know in lean
mode if you will for the past two or

three years and so
there hasn't been an official But

is this is definitely that's tied back to
MIT and the MIT business side is you know

involved in this but definitely our
goals is to get this into the clinic and

the really the big issue is matching
the market size with the research OK So

it turns out that if anyone's interested
in translation it's always informative to

really constantly talk with the business
folks because the business the market

might not fit the research interests and
so it turns out that where we were very

interested in liver fibrosis because
that's where most of our most robust that

is but the market for fibrosis is very
small and furthermore there really isn't

a therapeutic for fibrosis and so
if you were to catch fibrosis earlier.

So what the doctor there's no clinical
decisions us triggered and so

that actually is also
not a viable market but

yes we are actually
think about translating.