ACS Catalysis
Research Article
intramolecular peptide bond formation. Furthermore, we
explored the reaction mechanism by theoretical and exper-
imental approaches, which unveiled key advantages of the
multimetallic nature of the Zr-oxo cluster to the reactivity
observed. Initial results toward the intermolecular peptide
bond formation reaction are also presented.
minimal structural changes do not affect reactivity as both
substrates smoothly provide the desired cyclic adduct 2a,b in
>99% in MeOH. Using 1b, ethanol and 1-propanol also
provided 2b in >99% yield, while isopropanol resulted in only
∼60% yield, suggesting that smaller and linear alcohols are
preferred as reaction solvents (entries 5−8, Table 1), though
1
overlapped signals in crude H NMR hampered us form
RESULTS AND DISCUSSION
establishing whether the lower yield in isopropanol is derived
from lower conversion or simply from the lower mass recovery
in this reaction.
■
Reaction Optimization. To develop Zr-MOFs as catalysts
for the formation of peptide bonds, we have started our work
by studying the cyclization of dipeptides in the presence of Zr-
MOFs, following our previous successful strategy in which the
hydrolytic activity of Zr(IV)/Hf(IV)-polyoxometalate com-
plexes was repurposed to an amide bond formation reaction
upon rational adjustment of the reaction conditions.
the cyclization of glycylglycine (1a) to 2,5-diketopiperazine
2a) was evaluated in various solvents upon incubation of 1a
In general, the low mass recovery observed in some cases
seems to be related to the available uncoordinated Zr sites, and
the ability of the solvent to reverse the binding of substrates
and products to these sites, since lower mass recovery was
generally observed for MOF-808 and NU-1000 in comparison
with UiO-66. Similarly, lower mass recovery was also observed
when cyclization was carried out in solvents that poorly
solubilize 1a,b/2a,b (e.g., dioxane and toluene for UiO-66).
These trends strongly suggest that the interplay between the
available Zr sites and the ability of the solvent to reverse the
presumed coordination of substrates and products to these
sites plays a key role in the overall process efficiency. When
more Zr sites are available, more substrate/product molecules
bind to MOF, and if the solvent is not able to efficiently
reverse this trend, a lower mass recovery is observed. Such
hypothesis is also consistent with the better, but rather
intriguing performance of UiO-66 for the cyclization of 1a,b in
comparison with MOF-808 and NU-1000 MOFs, which
intrinsically have more uncoordinated Zr sites than UiO-66.
Together, these results show that the connectivity and missing-
linker defects alone do not guarantee higher reaction yield as
commonly indicated in the literature, since the adsorption of
substrates and products onto the MOF material and the ability
of the solvent to reverse it directly affects the overall mass
recovery after the reaction, subsequently impacting the
reaction yield. Interestingly, use of water as a reaction solvent
resulted in the lowest recoveries for MOF-808 and UiO-66,
suggesting solubility is likely not the only factor involved in
overcoming material adsorption.
Further optimization was performed by probing the effect of
temperature, concentration, and catalyst loading for the
reaction using UiO-66 in methanol. Lower temperature
decreased the reaction efficiency. At 70 °C, 2b yield slightly
°C or lower (entries 9−10, Table 1 and Table S2). In addition,
dilution or concentration of the reaction did not significantly
affect efficiency (entries 11−12, Table 1 and Table S3).
yields, although slower rates were observed (Table S4). Using
2 mol % of UiO-66, 2b was formed in 33% yield after 24 h.
However, a longer reaction time of 168 h provided a
quantitative yield, indicating that the catalyst is not deactivated
by side processes even after considerably long reaction time
(entries 13, Table 1). This robust profile is usual for Zr-MOFs
but is unparalleled to other metal catalysts for the amide bond
formation and further highlights the excellent catalytic
potential of Zr-MOFs for peptide bond formation.
17,18
Thus,
(
with different Zr-MOFs, which were previously shown to be
catalytically active for peptide bond hydrolysis (Table
4
6,48,49
17
1
).
Based on our previous work, we used DMSO as
a representative organic solvent to test three well known and
easily prepared Zr-MOFs with varying characteristics: (1)
MOF-808, a 6-connected MOF with 1,3,5-benzenetricarbox-
6
2
ylate linkers; (2) NU-1000, an 8-connected MOF with
6
3
1
,3,6,8-(p-benzoate)pyrene linkers; and (3) UiO-66, a
nominally 12-connected MOF with 1,4-benzenedicarboxylate
6
4
(
BDC) linkers. The presence of free carboxylic acid groups
in our substrate could result in adsorption of substrates or the
resulting products onto the MOF catalyst. Therefore, we have
employed a “washing step” after the reaction to ensure full
material recovery. Using MOF-808 as a representative
structure, we observed that stirring the crude reaction mixture
for 1 h with D O generally ensured a high recovery of material
2
(
>95%). We have standardized such step through all the
reaction discovery and optimization stages (Table S1).
In these initial reactions using DMSO as solvent, MOF-808
material observed for NU-1000 (<20%). MOF-808 formed 2a
in ∼50% yield, but a small fraction of glycine (Gly, 3a) was
also detected probably arising from peptide bond hydrolysis
with residual water present in the MOF structure and/or
solvent (Table 1). On the other hand, despite the lower
yield, UiO-66 cleanly converted 1a into 2a (33% yield).
Considering that MOF-808 and UiO-66 MOF allowed for an
easier recovery of substrates and products in our system,
further investigation focused on these MOFs only.
46
Following the initial screening, other solvents were probed
using MOF-808 and UiO-66, identifying alcohols as optimal
reaction solvents. Generally, MOF-808 adsorbed substrates
and products more strongly than UiO-66, as observed by the
mass recovery <75% for solvents other than MeOH and
DMSO for MOF-808. For UiO-66, only MeCN:H O, toluene,
2
Moreover, MOF-808 showed again a greater tendency than
UiO-66 to hydrolyze 1a, as evident from the results in solvents
containing water. On the other hand, UiO-66’s enhanced
selectivity toward 2a was once more observed for the reaction
in MeOH, which provided only desired product 2a and no
hydrolysis side reaction (entry 4, Table 1). To test other
alcohols, we used GlyAla (1b) instead of GlyGly (1a) to avoid
Control Experiments. Control experiments confirmed the
superiority of UiO-66 as a heterogeneous catalyst for
intramolecular peptide bond formation. In the absence of
reaction yields were <10% (Table S5). Notably, soluble
1
i
overlap between product and solvent peaks in H NMR. The
Zr( PrO) and a combination of ZrCl and BDC resulted in
4
4
7
649
ACS Catal. 2021, 11, 7647−7658