On-off transition and ultrafast decay of amino acid
luminescence driven by modulation of
Zohar A. Arnon,
Boris Yakimov, ...,
Davide Levy, Ehud
characterization of all
Lysine, a non-aromatic
amino acid, exhibited the
highest ﬂuorescent signal
between cysteine crystals
Ultrafast lifetime decay
conﬁrming governing role
Arnon et al., iScience 24,
July 23, 2021 ª2021
On-off transition and ultrafast decay
of amino acid luminescence driven
by modulation of supramolecular packing
Zohar A. Arnon,
Muhammad Nawaz Qaisrani,
and Ehud Gazit
Luminescence of biomolecules in the visible range of the spectrum has been
experimentally observed upon aggregation, contrary to their monomeric state.
However, the physical basis for this phenomenon is still elusive. Here, we system-
atically examine all coded amino acids to provide non-biased empirical insights.
Several amino acids, including non-aromatic, show intense visible luminescence.
Lysine crystals display the highest signal, whereas the very chemically similar
non-coded ornithine does not, implying a role for molecular packing rather than
the chemical characteristics. Furthermore, cysteine shows luminescence that is
indeed crystal packing dependent as repeated rearrangements between two
crystal structures result in a reversible on-off optical transition. In addition, ultra-
fast lifetime decay is experimentally validated, corroborating a recently raised
hypothesis regarding the governing role of np* states in the emission formation.
Collectively, our study supports that electronic interactions between non-ﬂuores-
cent, non-absorbing molecules at the monomeric state may result in reversible
optically active states by the formation of supramolecular ﬂuorophores.
In nature, proteins and peptides offer a broad range of characteristics and attributes, which provide an
extensive variety of mechanical, electrical, and optical properties. These are often derived from the phys-
icochemical properties of the amino acid building blocks that constitute the proteinaceous polymer. In
addition to the amino acid content, the arrangement of the protein in its environment, i.e., secondary, ter-
tiary, and quaternary structure, can dramatically inﬂuence the resulting physical properties. This arrange-
ment is affected by factors such as the solvent, temperature, and ionic strength. Recently, several studies
described the intrinsic luminescence in the visible range of protein assemblies, whereas the monomeric
protein in solution is not luminescent (Bhattacharya et al., 2017;Pinotsi et al., 2016). Consistently, it is
now evident that some amyloids, once assembled, exhibit ﬂuorescence that is not demonstrated in the
monomeric state in solution (Ardona et al., 2017;Chan et al., 2013;Del Mercato et al., 2007;Pinotsi
et al., 2013).
Monomeric amino acids were subjected to extensive research over the years. Recently, a great emphasis
was given to the ability of amino acids to serve as building blocks for the formation of various supramolec-
ular assemblies with attractive features (Guerin et al., 2018;Ji et al., 2019;Makam et al., 2019). It was shown
that single amino acids in their aggregated form can also produce a ﬂuorescent signal, which is not evident
in the monomeric state (Babar and Sarkar, 2017;Banerjee et al., 2020;Niyangoda et al., 2017;Shaham-Niv
et al., 2018). Numerous observations suggest that the optical phenomenon of aggregation-induced
intrinsic ﬂuorescence is broader than originally speculated, as it also applies to many other metabolites
and nucleic acids (Arnon et al., 2019;Berger et al., 2015;Chen et al., 2018;Lakowicz et al., 2001;Stephens
et al., 2020;Zou et al., 2002). Circumstantial evidence and argumentation suggested that supramolecular
packing is important. It was suggested that the hydrogen bonds and, in speciﬁc cases, the aromatic stack-
ing, can underlie these optical properties (Jong et al., 2019;Pinotsi et al., 2016). Yet, other views have sug-
gested impurities and oxidation as alternative mechanisms, ascribing no signiﬁcant role to the molecular
Department of Molecular
Biotechnology, George S.
Wise Faculty of Life Sciences,
Tel Aviv University, Tel Aviv
Faculty of Physics, M.V.
Lomonosov Moscow State
University, Moscow 119991
BLAVATNIK CENTER for
Drug Discovery, Metabolite
Medicine Division, Tel Aviv
University, Tel Aviv 6997801,
International Centre for
Theoretical Physics, Strada
Costiera, 11, 34151 Trieste,
Department of Physiology
and Pharmacology, Sackler
Faculty of Medicine, Tel Aviv
University, Tel Aviv University,
6997801 Tel Aviv, Israel
Institute for Regenerative
Medicine, I.M. Sechenov
Moscow State Medical
University, 119991 Moscow,
X-Ray Diffraction Lab,
Wolfson Applied Materials
Research Centre, Tel Aviv
University, Tel Aviv 6997801,
Department of Materials
Science and Engineering Iby
and Aladar Fleischman
Faculty of Engineering, Tel
Aviv University, Tel Aviv
These authors contributed
iScience 24, 102695, July 23, 2021 ª2021
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Figure 1. Fluorescence of the 20 coded amino acids
Images of all 20 amino acids, both brightﬁeld (left) and confocal ﬂuorescence at excitation wavelength of 405 nm (right). Amino acid samples as originally
obtained from Sigma-Aldrich are displayed in rows marked with OS (Original Sigma). Amino acids dissolved, reassembled, and lyophilized are displayed in
2iScience 24, 102695, July 23, 2021
arrangement (Tikhonova et al., 2018). Indeed, oxidation of aromatic moieties is known to result in the for-
mation of ﬂuorescent products. However, this cannot readily explain the emergence of intrinsic ﬂuores-
cence upon aggregation of non-aromatic species (Del Mercato et al., 2007;Jong et al., 2019;Stephens
et al., 2020). Overall, the key evidence to distinguish between these options and elucidate the mechanism
of intrinsic ﬂuorescence upon aggregation was so far missing. We believe that the reversible transitions be-
tween the luminescent and non-luminescent states upon controllable change of aggregate structure is
such an evidence. In this work, we aimed at designing and performing such an experiment.
The phenomenon of aggregation-dependent luminescence of proteins is a rapidly evolving ﬁeld (Ardona
et al., 2017;Arnon et al., 2019;Babar and Sarkar, 2017;Banerjee et al., 2020;Berger et al., 2015;Bhatta-
charya et al., 2017;Chan et al., 2013;Kong et al., 2019;Niyangoda et al., 2017;Pansieri et al., 2019;Pinotsi
et al, 2013,2016;Prasad et al., 2017;Shaham-Niv et al., 2018). Protein aggregates were shown to absorb
light at wavelengths above 300 nm and to exhibit a structure-speciﬁc ﬂuorescence in the visible range,
even in the absence of aromatic amino acids (Del Mercato et al., 2007). A plausible explanation for this phe-
nomenon is the formation of structure-speciﬁc supramolecular ﬂuorophores that are permissive to proton
transfer across hydrogen bonds (Grisanti et al., 2017;Pinotsi et al., 2016;Yamaguchi et al., 2009). Other hy-
potheses postulate that the ﬂuorescence is associated with electron-hole recombination due to charge
transfer between charged amino acids (Ardona et al., 2017;Kumar et al., 2020;Prasad et al., 2017)or
that deep-blue autoﬂuorescence stems from carbonyl double bonds of the protein backbone (Niyangoda
et al., 2017). Following these studies, we have decided to look into the intrinsic optical properties of single
amino acid and metabolite assemblies. Indeed, we have shown that adenine, phenylalanine, tyrosine, and
tryptophan exhibit autoﬂuorescence in the visible range upon self-assembly (Shaham-Niv et al., 2018). This
allows the detection of metabolite assemblies within living cells, without any use of external dyes that could
interfere with the ability to accurately model a given sample (Buell et al., 2010). Other derivatives of tyrosine
were also shown to exhibit luminescence upon aggregation (Ren et al., 2019). Yet, the underlying mecha-
nism of the autoﬂuorescence of these assemblies is controversial, as in all cases it can be attributed to the
aromatic moiety. In addition, the explanation of impurities as the source of ﬂuorescence is still a plausible
option. Here, we have endeavored to study the optical properties of all 20 coded amino acids in a system-
atic, non-biased approach. Each amino acid was optically characterized in two states: the original powder
form, as obtained from the commercial supplier, and after being dissolved in heated water and allowed to
cool down for recrystallization. Aiming to simplify the process of assembly and i n order to eliminate as many
factors as possible, we avoided altering solvents, salt concentrations, changing pH, etc. The systematic
evaluation of all amino acids, before and after recrystallization, provides mechanistic insights into the
broad phenomenon of assembly-dependent intrinsic ﬂuorescence.
First, we examined the optical properties of the crystalline powders of all 20 coded amino acids in the dried
powder form. In addition, each powder was dissolved in water at a high concentration (Table S1). In order to
completely dissolve the powders, the aqueous solution was heated to 90 C and vortexed until a clear,
transparent solution was achieved. The amino acid solutions were then cooled down gradually to room
temperature and incubated for several days to allow the formation of assemblies within the solution.
The samples were then lyophilized to attain a dried pow der of the amino acid assemblies. The optical prop-
erties of these powders were examined as well.
The brightﬁeld and confocal ﬂuorescence images obtained under excitation of 405 nm for all 20 amino
acids, as originally received (OS, original Sigma) and after dissolution and reassembly (DR, dissolved
and reassembled), as well as quantiﬁcation of the ﬂuorescence signals, are displayed in Figure 1 (see Fig-
ure S1 for spectra). For all samples, similar excitation and detection parameters were used. Based on the
current measurements there is no systematic trend between the optical properties and the polarity of the
amino acids. However, it does appear as though the charged amino acids tend to be characterized by
stronger ﬂuorescence, whereas those amino acids with hydrocarbon side chains with methyl groups
tend to exhibit very weak optical activity.
Figure 1. Continued
rows marked with DR (Dissolved and Reassembled). At the bottom of each amino acid panel are the normalized ﬂuorescence signals of the OS sample
(light blue) and the DR sample (dark blue).
See also Tables S1 and S2 and Figure S1.
iScience 24, 102695, July 23, 2021 3
We then explored the intense ﬂuorescence of L-lysine, which exhibited the brightest signal of all 20 amino
acids. We ﬁrst inquired whether the length of the amine residue chain of lysine, which has four carbons, plays
a role in the optical properties of the crystal. For this purpose, we explored the ﬂuorescence of three additional
molecules derived from lysine: L-ornithine, L-2,4-diaminobutyric acid, and DL-2,3-diaminopropionic acid,
which comprise 3, 2, and 1 side chain carbons, respectively (Figure 2A). The rationale of this experiment was
to alter the crystal packing by shortening the amine residue, thereby modulating the electronic interactions
by varying the distance between molecules, and to examine any effect on the resulting optical properties.
All four samples (including lysine) comprised small crystalline powders (Figure 2B). The ﬂuorescent signal
from each sample was obtained using confocal microscopy, similar to the amino acids shown in Figure 1.
Figure 2. Fluorescence of lysine and its derivatives
(A) The chemical scheme of lysine and its derivatives comprising shorter carbon chains.
(B and C) Microscopy images showing (B) brightﬁeld and (C) ﬂuorescence (excitation at 405 nm) analyses of the four powders. Pixel color represents the
intensity; black, non; blue, dim; white, strong.
(D) Crystalline structures of lysine and its derivatives as determined using PXRD.
See also Data S1 for additional crystallographic information.
4iScience 24, 102695, July 23, 2021
Intense ﬂuorescent signals were prominent in the molecules comprising 1-, 2-, and 4-carbon chains,
whereas in the 3-carbon chain molecule, ornithine, no visible signal was detected (Figure 2C). The powder
X-ray diffraction (PXRD) patterns of the samples were further examined in order to understand the molec-
ular arrangement of the crystals, which might provide insight into the favorable interactions allowing the
crystal ﬂuorescence (Figure 2D). The crystal structures of lysine, ornithine, and diaminopropionic acid
were previously published (Chiba et al., 1967;Mar
´et al., 2019;Williams et al., 2016). The cry stal struc-
ture of diaminobutyric acid was determined by thePXRD pattern obtained in this study CCDC Deposition #
1990651 (see STAR Methods). The results revealed the crystal packing of all four samples, allowing molec-
ular inspection of the optical phenomenon. A possible explanation lies in the type of hydrogen bond
network that is formed within the supramolecular structure. The NH
group of lysine hemihydrate and dia-
minobutyric acid, and the NH
of diaminopropionic acid, may interact with a water or chloride ions and
donate hydrogen bonds to the molecule complexed in the supramolecular structure. On the other
hand, in the case of ornithine, the NH
groups do not form hydrogen bonds that are donated from the
amine group to either a chloride ion or a water molecule in the complex. Thus, the differences in the supra-
molecular packing involving the lysine side chains, chloride ions, and water molecules can lead to
differences in the optical properties. We believe that the packing of the crystal is an important factor in
determining its optical behavior, much like the properties of microenvironments such as different solvents,
ionic strength, polarity, or molecular concentration could be signiﬁcant for structure-function attributes.
Following this line of evidence, we focused on possible interconnections between the optical and structural
properties of amino acid powders.
Next, each of the 20 amino acid samples was examined using PXRD to determine the crystalline structure.
The results are summarized in Table S2. Most amino acid crystals were composed of the same crystal struc-
ture in both the OS and the DR samples. Of the 20 amino acids, only cysteine and serine displayed different
powder diffraction between their respective OS and DR samples, with a single crystal packing observed in
each sample. However, both serine samples exhibited some level of ﬂuorescence, whereas the cysteine DR
sample did not show any detectable ﬂuorescence (Figure 1), making the differentiation between the
cysteine samples much more straightforward. In addition, the DR serine sample was found to comprise
a hemihydrate crystal packing (Table S2), adding another factor of complexity,whilebothcysteinecrystals
showed different arrangements of cysteine alone, with four molecules in both unit cells. Hence, we further
investigated the OS and DR samples of cysteine (Figures 3A and 3B), which showed a clear difference in the
ﬂuorescent signal (Figure 3C) and exhibited an orthorhombic and monoclinic packi ng, respectively (Figures
3Dand3E)(Harding and Long, 1968;Kerr and Ashmore, 1973). Since the process of attaining the OS pow-
der is unknown to us, we have decided to crystallize an orthorhombic cysteine crystal after dissolving the
amino acid in order to control the entire procedure. We were able to crystallize both crystal packings,
orthorhombic and monoclinic, depending on the crystallization temperature, 4 Cand25C, respectively,
as the only difference in the crystallization conditions. This allowed a reversible crystallization of both crys-
tal packings, exemplifying the signiﬁcance of the crystal packing to the optical properties of the crystal.
Thus, the ﬂuorescent OS powder (Figure 3F) was dissolved and recrystallized at 25 C to obtain the non-
ﬂuorescent monoclinic DR sample (Figure 3G) and then re-dissolved and recrystallized at 4 Ctoform
yet again the ﬂuorescent orthorhombic crystals (Figure 3H). To conﬁrm that the ﬂuorescence does not
stem from impurities, the non-ﬂuorescent DR sample was washed several times to remove any impurities
that may reside in the supernatant, before recrystallizing as orthorhombic crystals. Although the ﬂuores-
cent signal of cysteine is relatively low, a difference in the ﬂuorescent signal between the two crystal pack-
ings is clearly evident. A comparison of the two structures showsthat there is a subtle difference in the prox-
imity of the S-H bonds relative to each other. In the orthorhombic structure, the sulﬁde groups are packed
closer to each other and may facilitate charge transfer, as seen in previous studies (Del Mercato et al., 2007;
Pinotsi et al., 2016;Prasad et al., 2017). It is important to note that the overall macroscopic morphology is
dictated by other factors in addition to the packing, as evidenced by d ifferent morphologies for an identical
packing (compare Figures 3F–3H), and vice versa (compare Figures 3G–3H). For this reason, the crystal
packing was determined using PXRD for each sample and for every iteration. Morphologically, the recrys-
tallized orthorhombic structures (Figure 3H) are similar to the non-ﬂuorescent monoclinic structures (Fig-
ure 3G) and are dissimilar to the relatively large OS crystals (Figure 3F); hence, this afﬁrms that the differ-
ence in ﬂuorescence is not derived from any morphological discrepancy. As stated above, the crucial
experiment to prove the inﬂuence of crystal packing on intrinsic ﬂuorescent formation could be an obser-
vation of reversible transitions between the luminescent and non-luminescent states upon controllable
iScience 24, 102695, July 23, 2021 5
change of the crystal structure. We believe that the observed on-off switching of cysteine ﬂuorescence
upon changing its crystal packing (Figure 3) represents such an experiment.
In an attempt to explain the underlying mechanism of the self-assembly-induced ﬂuorescence of amino
acids, we tried to ﬁnd a common thread in the aggregation-induced ﬂuorescence related literature. First
principles quantum chemistry calculations of self-assembly-induced ﬂuorescence mainly deal with absorp-
tion spectra (Grisanti et al., 2017;Pinotsi et al., 2016;Stephens et al., 2020). Without novel absorption bands,
no novel ﬂuorescence bands may appear, and the key to the explanation of the presence of new emission
properties is the understanding of long-wavelength absorption formation. This paradigm was extended in
the recent work of Grisanti et al., where, using ab initio calculations, ﬂuorescence properties of model pep-
tides aggregates were assessed (Grisanti et al., 2020). Besides the prediction of the emission in the visible
spectral range, a conclusion was made on the time-resolved behavior of the aggregation-induced ﬂuores-
cence. Namely, the presence of an ultrafast ﬂuorescence decay and accompanying spectral diffusion, i.e., a
gradual ﬂuorescence emission redshift in time, on a sub-picosecond timescale, was described theoretically.
This result prompted us to search for experimental indication of an ultrafast component in a self-assembly-
induced ﬂuorescence system by means of the ultrafast spectroscopy. Taking into account the universal char-
acter of this emission, i.e., similarity of its spectral properties for a broad range of systems, we chose to work
with a known model system—ﬁbrillar structures that are formed as a result of phenylalanine self-assembly.
The ﬂuorescence of phenylalanine was previously characterized, and showed high ﬂuorescent signal, much
higher than the unfortunately too weak signal of cysteine, which was insufﬁcient for these measurements
(Shaham-Niv et al., 2018). Thus, by the example of the ﬁbrils made of phenylalanine, we aim at corroborating
the presence of an ultrafast ﬂuorescence decay of the self-assembly-induced ﬂuorescence.
Figure 3. Fluorescence of cysteine
For a Figure360 author presentation of Figure 3, see https://doi.org/10.1016/j.isci.2021.102695.
(A and B) Confocal microscopy images of the cysteine (A) OS sample and (B) DR sample.
(C) The normalized ﬂuorescence intensity of cysteine samples as a function of the emission wavelength (excitation at 405 nm).
(D and E) Crystalline structures of cysteine based on PXRD analysis showing (D) the orthorhombic packing of the OS sample and (E) the monoclinic packingof
the DR sample.
(F–H) Confocal imaging of the cysteine (F) OS sample, (G) DR sample, and (H) DR sample that was re-dissolved and recrystallized at different conditionsto
regain an orthorhombic packing. For confocal images, pixel color represents the intensity; black, non; blue, dim; white, strong.
See also Table S3 and Figure S2.
6iScience 24, 102695, July 23, 2021
The self-assembly of phenylalanine in aqueous solution was initiated by cooling a supersaturated solution
of phenylalanine (40 mg/ml) at 90 Cto20C, similar to previous work (Shaham-Niv et al., 2018). This leads
to the formation of elongated ﬁbrillary aggregates, as shown by bright-ﬁeld microscopy (Figure 4A). The
phenylalanine ﬁbrils exhibited relatively high ﬂuorescence emission in the visible spectral range, which is
absent in the monomeric state of phenylalanine, and were characterized by the wavelength-dependent
Stokes shift, i.e., the spectra measured at longer excitation wavelength exhibited red-shifted emission (Fig-
ure 4B). Such a behavior of the ﬂuorescence emission band is known as the red-edge excitation shift effect
and is characteristic for the self-assembly-induced ﬂuorescence of amino acids, peptides, and proteins (Ar-
non et al., 2019;Berger et al., 2015;Kumar et al., 2020). The presence of elongated ﬁbrillar structures was
detectable by ﬂuorescence lifetime imaging, without the use of any external dyes, only by using the auto-
ﬂuorescence signal of the sample (Figure 4C). The autoﬂuorescence decay curve obtained for the phenyl-
alanine ﬁbrils revealed the presence of 0.35- and 2.53-ns decay times after ﬁtting to a biexponential model
(Figure 4D). However, the temporal resolution, or the width of the instrument response function (IRF) of the
ﬂuorescence lifetime imaging setup, which employs the standard time-correlated single photon counting
technique, is 50 ps and does not allow for detecting any ultrafast decay components, which will be
completely masked by the 50-ps IRF (Rovnyagina et al, 2019,2020). Hence, we have utilized the ﬂuores-
cence up-conversion technique, which provides for sub-picosecond resolution of ﬂuorescence decay (for
more information, see STAR Methods).
Using ﬂuorescence up-conversion, it was observed that the phenylalanine ﬁbrils exhibited ultrafast decay
lifetimecomponentasfastas0.67ps(Figure 4E). The second decay component was characterized by a 12.6
ps decay. The presence of the sub-picosecond ﬂuorescence decay is in agreement with the calculation of
the recent published work (Grisanti et al., 2020), where the spectral diffusion and relaxation was observed
on a few hundred picoseconds timescale.
In the work of Grisanti et al., the ab initio nonadiabatic dynamics simulations were used to reveal charac-
teristic properties of the excited electronic states of model amyloid-like peptides. It was shown that a
visible (blue-green) ﬂuorescence could originate from the np* states localized on the amide groups. The
speciﬁc structure of amyloids gives rise to stabilization of some of these states, thus lowering the energy
gap between the ground and minimum np* state leading to the shift of absorption and ﬂuorescence to
the near UV-visible range. Grisanti and co-workers observed that, after excitation in the manifold of np*
states in 10–15 fs all trajectories from different np* states reached the lowest excited S
ing to Grisanti et al.) state. In the range of 15–40 fs after excitation UV ﬂuorescence from S
state is observed
andispeakedat3.3eV(370 nm). After that, on the timescale of 100–200 fs, internal vibration redistri-
bution occurs that leads to additional redshift of the emission spectrum, down to 2.0 eV (620 nm)
This evolution occurs on the timescale of 100 fs and is accompanied by the gradual shift of the ﬂuores-
cence emission from the UV to the blue-green region (Figure 4G). By setting the registration emission
wavelength to the blue-green region (in our case, to 500 nm [2.5 eV]), the ultrafast relaxation along the
potential curve shown in Figure 4F can be detected. This process is illustrated in the inset of Figure 4G
and corresponds to the ﬂuorescence decay curve presented in Figure 4E. The ultrafast relaxation occurs
on the timescale of 100 fs, corresponding to the non-radiative decay rate of 10
ever, lower than in the case of monomeric proteins, where the lack of stabilization of the np* states results in
the absence of near-UV and visible absorption and blue-green ﬂuorescence (Grisanti et al., 2020). More-
over, the presence of the ultrafast decay may result in low ﬂuorescence quantum yield of the aggrega-
tion-induced emission of proteins (0.01) (Niyangoda et al., 2017). Further relaxation from the lowest
state may then occur on a longer timescale owing to the stabilization of the geometry of
the aggregate, and this relaxation corresponds to the 1 to 2-ns ﬂuorescence lifetime, which was observed
for the self-assembly-induced ﬂuorescence in this paper (Figure 4D) and other previous works (Tikhonova
et al., 2018). Hence, by detecting the ultrafast decay of the ﬁbrillar structures formed as the result of phenyl-
alanine self-assembly, the model of Grisanti et al. can be extended frompeptides to amino acids, therefore
elucidating the origin of their enigmatic ﬂuorescence emission upon packing (Grisanti et al., 2020).
To conclude, in thiswork we have studiedthe optical properties of all20 coded amino acid powders. Eachpowder
was examinedas commercially obtained and also afterdissolution, reassembly, and lyophilization. It is surprising
iScience 24, 102695, July 23, 2021 7
that lysine exhibited the most intense ﬂuorescence, even though it has no aromatic moieties. Other charged
amino acidsshowed little to no ﬂuorescence. In addition, short-chain derivatives of lysinedisplayed no correlation
betweenthe chain length and their intrinsicﬂuorescence. This indicates that, in somecases, not only the chemical
identity ofthe monomeric moleculedictates the optical properties of the assembly, but alsothe supramolecular
arrangement within the assembly. To substantiate this notion, PXRD analysis together with confocal imaging re-
vealed that the orthorhombic crystal of cysteine is ﬂuorescent, whereas the monoclinic crystal is not. The recrys-
tallization of the non-ﬂuorescent monoclinic crystals into ﬂuorescent orthorhombic crystals conﬁrms that, in this
case, changing the crystal packing is sufﬁcient for conferring optical properties, and that the ﬂuorescence does
not stemfrom contaminations, impurities, or oxidation.It is important to note that we donot imply that impurities,
Figure 4. Ultrafast decay
(A) Chemical scheme and ﬁbrils of phenylalanine (macroscopic and microscopic photos).
(B) Fluorescence emission of phenylalanine assemblies at excitations between 370 and 420 nm.
(C) Autoﬂuorescence lifetime imaging of phenylalanine ﬁbrils. Excitation was performed in a two-photon regime at
(D) Fluorescence decay curve of the phenylalanine ﬁbrils measured with (sub)nanosecond resolution.
(E) Fluorescence decay curve of phenylalanine ﬁbrils measured using ﬂuorescence up-conversion technique. Excitation
and emission were set to 380 and 450 nm, respectively. The inset shows the parameters of ﬂuorescence decay obtained
using the biexponential decay model (D and E).
(F) Schematic decrease in the energy gap between the excited np* and ground state of the model ﬁbrils.
(G) Scheme of the shift of the aggregation-induced ﬂuorescence spectrum with time, demonstrating the presence of the
ultrafast decay and spectral migration.
8iScience 24, 102695, July 23, 2021
oxidation, or aromatic interactions are not valid mechanisms that couldexplain the ﬂuorescence in some cases.
However,in addition to those,supramolecularinteractions mayalso affect the opticalproperties of otherbiomol-
ecular assemblies, as unambiguously presented here, thus playing a key role in a phenomenon that could be ex-
plained byother known mechanisms. In the literature it was proposed that peptide aggregation-induced ﬂuores-
cence in the blue-green spectral range is due to (1) delocalization of electrons over a network of hydrogen bonds
(Shukla et al.,2004), (2) hydrogenbond-mediated interactionsbetween the amidegroups (Ye et al., 2017), (3) pro-
ton transfer across hydrogen bonds(Pinotsi et al., 2016),and (4) a decreasein the energy gap between the excited
and groundstates caused by the inﬂuence of the hydrogenbonds on the amide groupgeometry (Ye et al., 2017).
The importance and role of the structure for ﬂuorescenceformation in aggregates of non-aromatic peptideshas
been recently addressed using ab initio nonadiabatic dynamics simulations of the excited electronic states (Ye
et al., 2017). In addition, previous work has shown that low-energy optical excitations and subsequent ﬂuores-
cence can be induced by charge-transfer excitations (Jong et al., 2019;Mandal et al., 2019;Prasad et al., 2017).
Speciﬁcally, charge transfer excitations involving sulfuratoms of methionine and thepositively charged N termini
of amyloid aggregates were presented (Jong et al., 2019). In the case of cysteine demonstrated in this work, the
difference in the molecular packing within the crystal alters the distances between the sulfur atom and the N
termini of the neighboring molecule, which could affect the charge transfer potential. Indeed, these distances
are shorter for the orthorhombic crystal in comparison with the monoclinic packing (Figure S2). Overall, on the
basis of previous research, it can be summarized thatthe chromophore responsible forthe aggregation-induced
absorption and ﬂuorescence in the visible range is structure speciﬁc and can be formed either in the absence or
presence of aromatic moieties in peptides. Our results demonstrate that even the simplest and most thoroughly
investigated molecular systems, such as amino acids, can still serve as the basis for new, intriguing, and unknown
phenomena. The reversibility of cysteine ﬂuorescence serves as a strong evidence that the molecular arrange-
ment has a crucial role in the observed optical properties.
The prediction of ultrafast decay by Grisanti et al. was experimentally conﬁrmed in an aggregation-induced
ﬂuorescence model system of phenylalanine ﬁbrils. Speciﬁcally, the presence of the ultrafast decay of the
blue-green autoﬂuorescence is in agreement with the hypothesis regarding the governing role of amyloid
structure-stabilized np* states in the emission formation. Although the experimental data do not exclude
other hypotheses, its relevance to the theoretical calculations can be a step toward understanding the
origin of the blue-green emission in amyloids and other systems that recently attracts increased interest,
which appear as a result of peptides and amino acid self-assembly. Further understanding of the underlying
mechanisms could aid in harnessing the intrinsic properties of supramolecular polymers self-assembled by
simple and cost-effective building blocks to develop smart optoelectronic materials.
Limitations of the study
The current study aims to systematically explore, in a non-bias manner, the requirements for autoﬂuores-
cence upon aggregation. Nevertheless, it is limited only to the 20 coded amino acids. Although we
observed similar optical behavior for other small molecules, such as metabolites and nucleotides, further
research, beyond the scope of this study, is necessary to understand the fundamental principles of this phe-
nomenon. In addition, there are numerous other questions that could be asked and researched for
regarding these systems, such as the effect of racemic amino acids, different solvents, or ionic strength.
Questions like what is the ‘‘proper packing’’ for ﬂuorescence? What are the requirements? This is the
main enigma. Once we know the answer, we can potentially pre-design the optical properties of a material.
Although we did not ﬁnd a clear-cut answer to this question, we believe that this article is a big step in the
Detailed methods are provided in the online version of this paper and include the following:
dKEY RESOURCES TABLE
BData and code availability
BAmino acid reassembly
iScience 24, 102695, July 23, 2021 9
BPowder X-Ray diffraction
BPhenylalanine ﬁbrils sample preparation
BSteady-state ﬂuorescence measurements of phenylalanine ﬁbrils
BFluorescence lifetime microscopy (FLIM)
BSub-picosecond ﬂuorescence lifetime measurements
Supplemental information can be found online at https://doi.org/10.1016/j.isci.2021.102695.
Z.A.A. and T.K. contributed equally to this work. We thank our lab members for the fruitful discussions. This
work was supported by the Israeli National Nanotechnology Initiative and Helmsley Charitable Trust (E.G.),
ner Institute (Z.A.A.). The work was supported by the Ministry of Science and Higher Education of the
Russian Federation within the framework of state support for the creation and development of World-Class
Research Centers "Digital biodesign and personalized healthcare" No075-15-2020-926 (E.S.). We thank
members of the Gazit group for the helpful discussions.
Z.A.A., T.K., S.S.-N., and E.G. conceived and designed the experiments. Z.A.A., T.K., N.B., R.A., S.S.-N.,
P.M., M.N.Q., E.P., A.R., I.S., A.H., E.S., and D.L. planned and performed the experiments. Z.A.A., T.K.,
and E.G. wrote the manuscript. D.L. performed PXRD experiments and analysis. All authors discussed
the results, provided intellectual input and critical feedback, and commented on the manuscript.
DECLARATION OF INTERESTS
The authors declare no competing interests.
Revised: May 17, 2021
Accepted: June 4, 2021
Published: July 23, 2021
The following references appear in the supplemental information: Almlo
¨f et al., 1973;Coll et al., 1986;De-
rissen et al., 1968;Frey et al., 1973;Janczak and Luger, 1997;Kayushina, 1965;Khawas and Krishna Murti,
1969;King et al., 2012;Kistenmacher et al., 1974;Koetzle et al., 1973;Kvick et al., 1980;Lehmann et al., 1972;
Mazumdar et al., 1969;Mostad et al., 1973;Oda and Koyama, 1972;Ramanadham et al., 1973;Torii and
Iitaka, 1973;Torii and Iitaka, 1971;Torii and Iitaka, 1970;Verbist et al., 1972;Wilson et al., 2005;Yamada
et al., 2007.
¨f, J., Kvick, A
˚., and Thomas, J.O. (1973).
Hydrogen bond studies. 77. Electron density
distribution in a-glycine: X-N difference Fourier
synthesis vs ab initio calculations. J. Chem. Phys.
59, 3901–3906. https://doi.org/10.1063/1.1680574.
Altomare, A., Cuocci, C., Giacovazzo, C.,
Moliterni, A., Rizzi, R., Corriero, N., and
Falcicchio, A. (2013). EXPO2013: a kit of
tools for phasing crystal structures from
powder data. J. Appl. Crystallogr. 46, 1231–
Ardona, H.A.M., Draper, E.R., Citossi, F., Wallace,
M., Serpell, L.C., Adams, D.J., and Tovar, J.D.
(2017). Kinetically controlled coassembly of
multichromophoric peptide hydrogelators and
the impacts on energy transport. J. Am. Chem.
Soc. 139, 8685–8692. https://doi.org/10.1021/
Arnon, Z.A., Berger, O., Aizen, R., Hannes, K.,
Brown, N., Shimon, L.J.W., and Gazit, E. (2019).
Coassembly of complementary peptide nucleic
acid into crystalline structures by microﬂuidics.
Small Methods 3, 1900179. https://doi.org/10.
Babar, D.G., and Sarkar, S. (2017). Self-assembled
nanotubes from single ﬂuorescent amino acid.
Appl. Nanosci. 7, 101–107. https://doi.org/10.
Banerjee, P., Pyne, A., and Sarkar, N. (2020).
Understanding the self-assembling
behavior of biological building block
molecules: a spectroscopic and
microscopic approach. J. Phys. Chem. B 124,
Berger, O., Adler-Abramovich, L., Levy-Sakin,
M., Grunwald, A., Liebes-Peer, Y., Bachar,
Schwartz, T., et al. (2015). Light-emitting
self-assembled peptide nucleic acids
10 iScience 24, 102695, July 23, 2021
exhibit both stacking interactions and
Watson-Crick base pairing. Nat.
Nanotechnol. 10, 353–360. https://doi.org/10.
Bhattacharya, A., Bhowmik, S., Singh, A.K.,
Kodgire, P., Das, A.K., and Mukherjee, T.K. (2017).
Direct evidence of intrinsic blue ﬂuorescence
from oligomeric interfaces of human serum
albumin. Langmuir 33, 10606–10615. https://doi.
Buell, A.K., Dobson, C.M., Knowles, T.P.J., and
Welland, M.E. (2010). Interactions between
amyloidophilic dyes and their relevance to
studies of amyloid inhibitors. Biophys. J. 99,
Chan, F.T.S., Kaminski Schierle, G.S., Kumita, J.R.,
Bertoncini, C.W., Dobson, C.M., and Kaminski,
C.F. (2013). Protein amyloids develop an intrinsic
ﬂuorescence signature during aggregation.
Analyst 138, 2156–2162. https://doi.org/10.1039/
Chen, X., Luo, W., Ma, H., Peng, Q., Yuan, W.Z.,
and Zhang, Y. (2018). Prevalent intrinsic emission
from nonaromatic amino acids and poly(amino
acids). Sci. China Chem. 61, 351–359. https://doi.
Chiba, A., Ueki, T., Ashida, T., Sasada, Y., and
Kakudo, M. (1967). The crystal structure of L-
ornithine hydrochloride. Acta Crystallogr. 22,
Coll, M., Solans, X., Font-Altaba, M., and
Subirana, J.A. (1986). Structure of L-leucine: a
redetermination. Acta Crystallogr. C Cryst. Struct.
Commun. 42, 599–601. https://doi.org/10.1107/
Del Mercato, L.L., Pompa, P.P., Maruccio, G.,
Della Torre, A., Sabella, S., Tamburro, A.M.,
Cingolani, R., and Rinaldi, R. (2007). Charge
transport and intrinsic ﬂuorescence in amyloid-
like ﬁbrils. Proc. Natl. Acad. Sci. U S A 104, 18019–
Derissen, J.L., Endeman, H.J., and Peerdeman,
A.F. (1968). The crystal and molecular structure of
L-aspartic acid. Acta Crystallogr. B Struct.
Crystallogr. Cryst. Chem. 24, 1349–1354. https://
Frey, M.N., Lehmann, M.S., Koetzle, T.F., and
Hamilton, W.C. (1973). Precision neutron
diffraction structure determination of protein and
nucleic acid components. XI. Molecular
conﬁguration and hydrogen bonding of serine in
the crystalline amino acids L-serine monohydrate
and DL-serine. Acta Crystallogr. B Struct.
Crystallogr. Cryst. Chem. 29, 876–884. https://
Grisanti, L., Pinotsi, D., Gebauer, R., Kaminski
Schierle, G.S., and Hassanali, A.A. (2017). A
computational study on how structure inﬂuences
the optical properties in model crystal structures
of amyloid ﬁbrils. Phys. Chem. Chem. Phys. 19,
Grisanti, L., Sapunar, M., Hassanali, A., and
c, N. (2020). Toward understanding optical
properties of amyloids: a reaction path and
nonadiabatic dynamics study. J. Am. Chem. Soc.
142, 18042–18049. https://doi.org/10.1021/jacs.
Guerin, S., Stapleton, A., Chovan, D., Mouras, R.,
Gleeson, M., McKeown, C., Noor, M.R., Silien, C.,
Rhen, F.M.F.F., Kholkin, A.L.L., et al. (2018).
Control of piezoelectricity in amino acids by
supramolecular packing. Nat. Mater. 17, 180–186.
Harding, M.M., and Long, H.A. (1968). The crystal
and molecular structure of L-cysteine. Acta
Crystallogr. B Struct. Crystallogr. Cryst. Chem. 24,
Janczak, J., and Luger, P. (1997). L-proline
monohydrate at 100 K. Acta Crystallogr. C Cryst.
Struct. Commun. 53, 1954–1956. https://doi.org/
Ji, W., Xue, B., Arnon, Z.A.Z.A., Yuan, H., Bera, S.,
Li, Q., Zaguri, D., Reynolds, N.P.N.P., Li, H., Chen,
Y., et al. (2019). Rigid tightly packed amino acid
crystals as functional supramolecular materials.
ACS Nano 13, 14477–14485. https://doi.org/10.
Jong, K.H., Azar, Y.T., Grisanti, L., Stephens, A.D.,
Jones, S.T.E.E., Credgington, D., Kaminski
Schierle, G.S., and Hassanali, A. (2019). Low
energy optical excitations as an indicator of
structural changes initiated at the termini of
amyloid proteins. Phys. Chem. Chem. Phys. 21,
Kayushina, R.L.V.B.R. (1965). Opredepenie
strukturi L-prolina. Kristallograﬁya 10, 833–844.
Kerr, K.A., and Ashmore, J.P. (1973). Structure and
conformation of orthorhombic L-cysteine. Acta
Crystallogr. B Struct. Crystallogr. Cryst. Chem. 29,
Khawas, B., and Krishna Murti, G.S.R. (1969). On
the unit-cell dimensions and space group of L-
tyrosine and L-tryptophane. Acta Crystallogr. B
Struct. Crystallogr. Cryst. Chem. 25, 1006–1009.
King, M.D., Blanton, T.N., and Korter, T.M. (2012).
Revealing the true crystal structure of L-
phenylalanine using solid-state density functional
theory. Phys. Chem. Chem. Phys. 14, 1113–1116.
Kistenmacher, T.J., Rand, G.A., and Marsh, R.E.
(1974). Reﬁnements of the crystal structures of DL-
serine and anhydrous L-serine. Acta Crystallogr. B
Struct. Crystallogr. Cryst. Chem. 30, 2573–2578.
Koetzle, T.F., Frey, M.N., Lehmann, M.S., and
Hamilton, W.C. (1973). Precision neutron
diffraction structure determination of protein and
nucleic acid components. XIII. Molecular and
crystal structure of the amino acid L-glutamine.
Acta Crystallogr. B Struct. Crystallogr. Cryst.
Chem. 29, 2571–2575. https://doi.org/10.1107/
Kong, J., Wang, Y., Qi, W., Su, R., and He, Z.
(2019). Photo- and aromatic stacking-induced
green emissive peptidyl nanoparticles for cell
imaging and monitoring of nucleic acid delivery.
ACS Appl. Mater. Interfaces 11, 15401–15410.
Kumar, A., Ahari, D., Priyadarshi, A., Ziauddin
Ansari, M., and Swaminathan, R. (2020). Weak
intrinsic luminescence in monomeric proteins
arising from charge recombination. J. Phys.
Chem. B 124, 2731–2746. https://doi.org/10.
˚., Canning, W.M., Koetzle, T.F., and
Williams, G.J.B. (1980). An experimental study of
the inﬂuence of temperature on a hydrogen-
bonded system: the crystal structure of g-glycine
at 83 K and 298 K by neutron diffraction. Acta
Crystallogr. B Struct. Crystallogr. Cryst. Chem. 36,
Lakowicz, J.R., Shen, B., Gryczynski, Z., D’Auria,
S., and Gryczynski, I. (2001). Intrinsic ﬂuorescence
from DNA can be enhanced by metallic particles.
Biochem. Biophys. Res. Commun. 286, 875–879.
Lehmann, M.S., Koetzle, T.F., and Hamilton, W.C.
(1972). Precision neutron diffraction structure
determination of protein and nucleic acid
components. VIII: the crystal and molecular
structure of the b-form of the amino acidl-
glutamic acid. J. Cryst. Mol. Struct. 2, 225–233.
Makam, P., Yamijala, S.S.R.K.C., Tao, K., Shimon,
L.J.W., Eisenberg, D.S., Sawaya, M.R., Wong,
B.M., and Gazit, E. (2019). Non-proteinaceous
hydrolase comprised of a phenylalanine metallo-
supramolecular amyloid-like structure. Nat. Catal.
2, 977–985. https://doi.org/10.1038/s41929-019-
Mandal, I., Manna, S., and Venkatramani, R.
(2019). UV–Visible lysine–glutamate dimer
excitations in protein charge transfer spectra:
TDDFT descriptions using an optimally tuned
CAM-B3LYP functional. J. Phys. Chem. B 123,
´, M., Ger
Soral, M., Moncol, J.,
s, D., Kolarovi
c, A., and Jakubec, P. (2019).
Crystallization does it all: an alternative strategy
for stereoselective aza-henry reaction. Org. Lett.
21, 4580–4584. https://doi.org/10.1021/acs.
Mazumdar, S.K., Venkatesan, K., Mez, H.-C., and
Donohue, J. (1969). The crystal structure of l-
arginine hydrochloride. Z. fu
¨r Krist 130, 328–339.
Mostad, A., Rømming, C., Graver, H., Husebye,
S., Klæboe, P., and Swahn, C.-G. (1973). Crystal
structure of DL-tyrosine. Acta Chem. Scand. 27,
Niyangoda, C., Miti, T., Breydo, L., Uversky, V.,
and Muschol, M. (2017). Carbonyl-based blue
autoﬂuorescence of proteins and amino acids.
PLoS One 12, e0176983. https://doi.org/10.1371/
Oda, K., and Koyama, H. (1972). A reﬁnement of
the crystal structure of histidine hydrochloride
monohydrate. Acta Crystallogr. B 28, 639–642.
Pansieri, J., Josserand, V., Lee, S.J., Rongier, A.,
Imbert, D., Sallanon, M.M., Ko
¨vari, E., Dane, T.G.,
Vendrely, C., Chaix-Pluchery, O., et al. (2019).
iScience 24, 102695, July 23, 2021 11
properties of amyloid ﬁbrils shed light on
amyloidogenesis. Nat. Photon. 13, 473–479.
Pinotsi, D., Buell, A.K., Dobson, C.M., Kaminski
Schierle, G.S., and Kaminski, C.F. (2013). A label-
free, quantitative assay of amyloid ﬁbril growth
based on intrinsic ﬂuorescence. Chembiochem
14, 846–850. https://doi.org/10.1002/cbic.
Pinotsi, D., Grisanti, L., Mahou, P., Gebauer, R.,
Kaminski, C.F., Hassanali, A., and Kaminski
Schierle, G.S. (2016). Proton transfer and
structure-speciﬁc ﬂuorescence in hydrogen
bond-rich protein structures. J. Am. Chem. Soc.
138, 3046–3057. https://doi.org/10.1021/jacs.
Prasad, S., Mandal, I., Singh, S., Paul, A., Mandal,
B., Venkatramani, R., and Swaminathan, R. (2017).
Near UV-Visible electronic absorption originating
from charged amino acids in a monomeric
protein. Chem. Sci. 8, 5416–5433.
Ramanadham, M., Sikka, S.K., and Chidambaram,
R. (1973). Structure determination of Ls-threonine
by neutron diffraction. Pramana 1, 247–259.
Ren, X., Zou, Q., Yuan, C., Chang, R., Xing, R., and
Yan, X. (2019). The dominant role of oxygen in
modulating the chemical evolution pathways of
tyrosine in peptides: dityrosine or melanin.
Angew. Chem. Int. Ed. Engl. 58, 5872–5876.
Rovnyagina, N.R., Budylin, G.S., Vainer, Y.G.,
Tikhonova, T.N., Vasin, S.L., Yakovlev, A.A.,
Kompanets, V.O., Chekalin, S.V., Priezzhev, A.V.,
and Shirshin, E.A. (2020). Fluorescence lifetime
and intensity of thioﬂavin T as reporters of
different ﬁbrillation stages: insights obtained
from ﬂuorescence up-conversion and particle size
distribution measurements. Int. J. Mol. Sci. 21,
Rovnyagina, N.R., Tikhonova, T.N., Kompanets,
V.O., Sluchanko, N.N., Tugaeva, K.V., Chekalin,
S.V., Fadeev, V.V., Lademann, J., Darvin, M.E.,
and Shirshin, E.A. (2019). Free and bound
Thioﬂavin T molecules with ultrafast relaxation:
implications for assessment of protein binding
and aggregation. Laser Phys. Lett. 16, 075601.
Shaham-Niv, S., Arnon, Z.A., Sade, D.,
Lichtenstein, A., Shirshin, E.A., Kolusheva, S., and
Gazit, E. (2018). Intrinsic ﬂuorescence of
metabolite amyloids allows label-free monitoring
of their formation and dynamics in live cells.
Angew. Chem. Int. Ed. Engl. 57, 12444–12447.
Shukla, A., Mukherjee, S., Sharma, S., Agrawal, V.,
Radha Kishan, K., Guptasarma, P., Kishan, K.V.R.,
and Guptasarma, P. (2004). A novel UV laser-
induced visible blue radiation from protein
crystals and aggregates: scattering artifacts or
ﬂuorescence transitions of peptide electrons
delocalized through hydrogen bonding? Arch.
Biochem. Biophys. 428, 144–153.
Stephens, A.D., Qaisrani, M.N., Ruggiero, M.T.,
Jones, S.T.E., Poli, E., Bond, A.D., Woodhams,
P.J., Kleist, E.M., Grisanti, L., Gebauer, R., et al.
(2020). Intrinsic Fluorescence in Non-aromatic
Peptide Structures Is Induced by Collective
Vibrations, Charge Reorganisation and Short
Hydrogen Bonds, as Shown in a New Glutamine-
Related Structure (bioRxiv). https://doi.org/10.
Tikhonova, T.N., Rovnyagina, N.R., Zherebker,
A.Y., Sluchanko, N.N., Rubekina, A.A., Orekhov,
A.S., Nikolaev, E.N., Fadeev, V.V., Uversky, V.N.,
and Shirshin, E.A. (2018). Dissection of the deep-
blue autoﬂuorescence changes accompanying
amyloid ﬁbrillation. Arch. Biochem. Biophys. 651,
Toby, B.H., and Von Dreele, R.B. (2013). GSAS-II:
the genesis of a modern open-source all purpose
crystallography software package. J. Appl.
Crystallogr. 46, 544–549. https://doi.org/10.1107/
Torii, K., and Iitaka, Y. (1973). Crystal structures
and molecular conformations of L-methionine
and L-norleucine. Acta Crystallogr. B Struct.
Crystallogr. Cryst. Chem. 29, 2799–2807. https://
Torii, K., and Iitaka, Y. (1971). The crystal structure
of L-isoleucine. Acta Crystallogr. B Struct.
Crystallogr. Cryst. Chem. 27, 1317–1326. https://
Torii, K., and Iitaka, Y. (1970). The crystal structure
of L-valine. Acta Crystallogr. B 26, 1317–1326.
Verbist, J.J., Lehmann, M.S., Koetzle, T.F., and
Hamilton, W.C. (1972). Precision neutron
diffraction structure determination of protein and
nucleic acid components. VI. The crystal and
molecular structure of the amino acid L-
asparagine monohydrate. Acta Crystallogr. B
Struct. Crystallogr. Cryst. Chem. 28, 3006–3013.
Williams, P.A., Hughes, C.E., Martin, J.,
Courvoisier, E., Buanz, A.B.M., Gaisford, S., and
Harris, K.D.M. (2016). Understanding the solid-
state hydration behavior of a common amino
acid: identiﬁcation, structural characterization,
and hydration/dehydration processes of new
hydrate phases of l-lysine. J. Phys. Chem. C 120,
Wilson, C.C., Myles, D., Ghosh, M., Johnson, L.N.,
and Wang, W. (2005). Neutron diffraction
investigations of l- and d-alanine at different
temperatures: the search for structural evidence
for parity violation. New J. Chem. 29, 1318.
Yamada, K., Hashizume, D., Shimizu, T., and
Yokoyama, S. (2007). L-Asparagine. Acta
Crystallogr. E Struct. Rep. Online 63, 3802–3803.
Yamaguchi, S., Kamikubo, H., Kurihara, K., Kuroki,
R., Niimura, N., Shimizu, N., Yamazaki, Y., and
Kataoka, M. (2009). Low-barrier hydrogen bond in
photoactive yellow protein. Proc. Natl. Acad. Sci.
USA106, 440–444. https://doi.org/10.1073/
Ye, R., Liu, Y., Zhang, H., Su, H., Zhang, Y., Xu, L.,
Hu, R., Kwok, R.T.K., Wong, K.S., Lam, J.W.Y.,
et al. (2017). Non-conventional ﬂuorescent
biogenic and synthetic polymers without
aromatic rings. Polym. Chem. 8, 1722–1727.
Zou, L., Harkey, M.R., and Henderson, G.L. (2002).
Effects of intrinsic ﬂuorescence and quenching on
ﬂuorescence-based screening of natural
products. Phytomedicine 9, 263–267. https://doi.
12 iScience 24, 102695, July 23, 2021
KEY RESOURCES TABLE
Further information and requests for resources and materials should be directed to and will be fulﬁlled by
Direct material requests to lead contact, Ehud Gazit.
Data and code availability
The published article includes all datasets generated or analyzed during this study. Crystallographic Infor-
mation File (CIF) for the crystal structure of L-2,4-diaminobutyric acid can be found in the CCDC database
(CCDC Deposition # 1990651).
REAGENT or RESOURCE SOURCE IDENTIFIER
Chemicals, peptides, and recombinant proteins
L-Alanine Sigma-Aldrich A7627
L-Arginine monohydrochloride Sigma-Aldrich 11039
L-Aspargine Sigma-Aldrich A-0884
L-Aspartic acid Sigma-Aldrich A9256
L-Cysteine Sigma-Aldrich 168179
L-Glutamine Sigma-Aldrich G-3126
L-Glutamic acid Sigma-Aldrich G1251
L-Glycine Sigma-Aldrich G7126
L-Histidine monohydrochloride monohydrate Sigma-Aldrich H8125
L-Isoleucine Sigma-Aldrich I2752
L-Leucine Sigma-Aldrich L8000
L-Lysine Sigma-Aldrich L5501
L-Methionine Sigma-Aldrich M9625
L-Phenylalanine Sigma-Aldrich P2126
L-Proline Sigma-Aldrich P-0380
L-Serine Sigma-Aldrich S4311
L-Threonine Sigma-Aldrich T8625
L-Tryptophan Sigma-Aldrich T0254
L-Tyrosine Sigma-Aldrich T3754
L-Valine Sigma-Aldrich V0500
Crystal structure of L-2,4-diaminobutyric acid This Paper; Cambridge Crystallographic Data
Software and algorithms
GraphPad Prism 9.0.1 GraphPad https://www.graphpad.com/
ImageJ 1.52 NIH http://imagej.nih.gov.ij
Inkscape 0.92 Inkscape.org https://www.inkscape.org
Mercury 4.1.3 Cambridge Crystallographic Data Centre http://www.ccdc.cam.ac.uk/mercury/
ChemBioDraw Ultra 14.0 PerkinElmer https://scistore.cambridgesoft.com/
iScience 24, 102695, July 23, 2021 13
Amino acid reassembly
The original samples from Sigma-Aldrich were dissolved in double distilled water, at a relatively high
concentration, depending on the water solubility, as detailed in Table S1. The samples were then heated
to 90 C and vortexed to obtain a clear, transparent solution. The solutions were then incubated at room
temperature for a week to allow self-assembly and crystallization. The samples were then lyophilized to re-
move the water and attain a dry crystalline powder.
All confocal images were taken using a Leica SP8 Lightning confocal microscope with a Leica Application
Suite X (LAS X) software. The samples were excited using a 405 nm laser, with laser intensity set to 50% and
the gain set to 500 for all images. The emission range was set between 415 nm to 600 nm. All images were
taken at a magniﬁcation of 5X.
Images were analyzed using Image J. Since we use confocal microscopy, we can assume that the depth (Z)
is very small in comparison to the width and height (X and Y). Thus, in these images we can consider the
areas rather than the volumes. For each image, we divided the number of ‘‘ﬂuorescent’’ pixels by the num-
ber of ‘‘black’’ pixels in the corresponding brightﬁeld image. The threshold for ‘‘ﬂuorescent’’ and ‘‘black’’
pixels were identical for all the images in each experimental set.
Powder X-Ray diffraction
X-ray diffraction was collected using a Bruker D8 Discover diffractometer with LYNXEYE EX linear position
detector. The diffraction pattern to analyze the amino acid structure was performed in a classical w-wBragg-
Brentano setup. The diffraction patterns were corroborated based on the reported phases in the PDF-4-
organics-2019 database. As the crystal structure of L-2,4-diaminobutyric acid was not reported in the
literature, a full crystal structure determination was performed (CCDC Deposition # 1990651; Data S1).
The crystalized powder was placed in a 0.7 mm quartz capillary and a full diffraction pattern was collected
between 2 and 50 ,step0.02A
˚. The capillary setup was employed: Go
¨bels mirrors to obtain a parallel
beam, rotating capillary holder. The crystallographic structure determination was performed using the
EXPO2014 software (Altomare et al., 2013). These EXPO2104 features used were cell indexing (N-TREOR09
algorithm) and the Simulated Annealing Method.
The solution with the lowest cost function was used as model to perform further crystal reﬁnement on the
structure using the GSASII software (Toby and Von Dreele, 2013). The ﬁnal error indexes were: wR=10.9%
Both cysteine crystal packings were crystallized in double distilled water, at an amino acid concentration of
200 mg/ml. Monoclinic packing was attained as described in the ‘‘Amino Acid Reassembly’’ section above.
Orthorhombic packing was attained by incubating the solution over night at 4 C.
Phenylalanine ﬁbrils sample preparation
The sample preparation protocol was analogous to the procedures previously used to create self-assem-
bled phenylalanine -aggregates as previously described (Shaham-Niv et al., 2018) using the heat-cool tech-
nique. Brieﬂy, L-phenylalanine (PanreacApplichem, CAS 63-91-2, no additional puriﬁcation) in a concentra-
tion of 40 mg/ml was dissolved in distilled water (Millipore-Q) at 90 C (temperature was controlled by the
thermostat Qpod 2e, Quantum Northwest, USA) and was stirred for 1 hour using magnetic stirring for com-
plete dissolving. To obtain self-assembled aggregates, the heated phenylalanine solution was cooled to a
room temperature of 23-25 C under normal conditions in cuvettes or on glass slides, depending on the
type of the measurement being performed.
Steady-state ﬂuorescence measurements of phenylalanine ﬁbrils
Steady-state ﬂuorescence measurements were performed using the FluoroMax-4 spectroﬂuorometer
(HORIBA Jobin Yvon, Japan). Excitation-emission matrices were measured in the 400-750 nm emission
14 iScience 24, 102695, July 23, 2021
range with 1 nm step; the excitation wavelength was varied in the 370-420 nm range with a 10 nm step. The
spectral widths of the excitation and emission slits were set to 5 nm. The measurements were carried out in
a quartz cuvette with an optical path of 1 cm. The measurements were carried out for samples cooled to
Fluorescence lifetime microscopy (FLIM)
Fluorescence lifetime imaging microscopy (FLIM) with multiphoton excitation was performed using a
custom-build multiphoton multimodal microscopy setup. Femtosecond optical parametric oscillator
TOPOL-1050-C (Avesta, Russia), providing excitation by signal wave in the 680-1000 nm range, was used
as an excitation source. The pulse width of exciting radiation was 150 fs, with frequency of 80 MHz,
average power at the excitation wavelength on the sample was 1 mW. Scanning overthesamplewasper-
formed using DSC-120 scan head (Becker&Hickl, Germany). Imaging was performed using oil immersion
Plan Apochromat 603objective with NA=1.4 (CFI Plan Apochromat Lambda 603,Nikon,Japan).Fluores-
cence decay curves were detected using hybrid GaAsP detector HPM-100-40 (Becker&Hickl, Germany) with
sensitivity in 250-720 nm range and instrument response function characteristic time-width of 120 ps. To cut
off the excitation, a 680 nm short-pass dielectric ﬁlter was used. Both autoﬂuorescence of phenylalanine
-ﬁbers and the thioﬂavin T ﬂuorescence signals were excited at 730 nm.
Fluorescence decay curves were ﬁtted using SPCImage 8.3 software (Becker&Hickl, Germany) after spatial
binning (bin size was equal to 5 for ThT-ﬂuorescence measurements and 10 for autoﬂuorescence measure-
ments) by bi-exponential decay law with respect to instrument response function. Average ﬂuorescence
lifetime was calculated as t
), where a
are the amplitudes and lifetimes
obtained from ﬁt.
Sub-picosecond ﬂuorescence lifetime measurements
Time-resolved ﬂuorescence emission measurementsin the sub-picosecond time range were carried out us-
ing a commercially available femtosecond ﬂuorescence spectrometry system FOG100 (CDP Systems,
Russia). The samples were excited by 100 fs pulses at 380 nm with a frequency of 80 MHz (second harmonic
of Ti:Sapphire oscillator Mai-Tai, Spectra Physics, USA). The ﬂuorescence signal from the sample was
focused on a 0.5 mm b-barium borate crystal alongside the fundamental beam (80 fs, 760 nm), acting as
a gate pulse for the frequency up-conversion. The gate pulse was delayed by an automatically controlled
delay stage. The upconverted light was focused onto the entrance slit of the double monochromator (spec-
tral resolution < 1.5 nm) and was detected by a photomultiplier tube. The reproducibility of the measure-
ment was checked by 10 times, measuring decay trace. Special rotation cuvette unit was used to avoid pho-
todegradation of the sample.
Similar to steady-state ﬂuorescence and microscopy measurements, a phenylalanine sample in a liquid
state, heated to a temperature of 90 C, was poured into a cuvette and was cooled down to room temper-
ature for 15-20 minutes. The measurements were carried out when the sample was turbid. Fluorescence
decay curves were ﬁtted by biexponential decay law with respect to instrument response function ﬁtted
as Gaussian function. Data analysis on the sub-picosecond decay curves were performed using custom-
built Python scripts using LmFit, Matplotlib, Numpy, Pandas libraries.
[CCDC: 1990651 contains the supplementary crystallographic data for this paper. These data can be ob-
tained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/
iScience 24, 102695, July 23, 2021 15