solution. However, during the course of peptide synthesis the
rejection of the MeO-PEG-peptide is expected to increase up
to ∼100% due to the increasing molecular weight of the
growing peptide chain and its bulkier structure, and so in fact
product losses will be considerably less.
and nonbranched, and should permeate easily through the
membrane. DIC is used only in combination with HOBt as
racemisation suppressor, and so it is important to examine the
rejection of HOBt. There are two types of HOBt available on
the market. One is dry, crystalline flake HOBt, the other is a
wet powder, HOBt ·H2O. As shown in entries 11 and 12, the
moisture content seems to considerably affect the rejection of
HOBt by the membrane. The other two activators, PyBOP and
HBTU, do not require addition of HOBt41 because the latest
breaks off as a product from these activators during the coupling
reaction. The rejection of the break-off product is expected to
be low, since it is free of moisture. Finally the rejection of
HMPA linker (entry 15, Table 1) was also investigated in order
to evaluate diafiltration as an option to purify MeO-PEG-HMPA
after the attachment of linker onto MeO-PEG-NH2. Although
the rejection of HMPA is relatively low (39%) with 99%
rejection of MeO-PEG-HMPA, the product losses will still be
∼10% during the separation step, and it was decided to purify
MeO-PEG-HMPA by a precipitation method (see Experimental
Section).
The solvent flux through this membrane was stable, ∼27-30
L·m-2 ·h-1. With the membrane area of ∼0.011 m2, a reason-
able permeate flow was achieved. On the basis of the rejection
data and material balance equations for the diafiltration, a
mathematical simulation was performed to estimate the wash
solvent volume required for removal of excess reagents and
reaction byproducts after each reaction step (Figure 6). This
simulation for the coupling reaction was based on 2 equiv of
starting reagents per 1 equiv MeO-PEG-HMPA and assumes
100% conversion after each coupling step. As intuitively
expected, the two reagents with the highest rejections, the
activator PyBOP and the protected Fmoc-Arg(Boc)2 amino acid,
require the largest wash solvent volume (∼10 volumes per
starting volume) for removal to less than 0.05 equiv. Figure 6
suggests that, after 10 volumes of wash solvent per starting
volume, the washout becomes less efficient, and since most
reagents remaining at this point are below 0.03 equiv, simula-
tions were limited to 10-12 volumes. For the deprotection
reaction, despite the high excess of piperidine employed (∼40
equiv) due to its low rejection of ∼5% (entry 16, Table 1),
10-12 wash solvent volumes should be sufficient for its almost
complete removal to less than 0.01 equiv. Other chemicals with
low rejection such as DIC and HOBt should also be readily
removed with 10 volumes of diafiltration solvent.
The membrane rejection of a number of Fmoc-amino acids
was determined, covering a wide range of propertiessthe lowest
MW amino acid and the highest MW amino acid, acidic, basic,
and hydrophobic amino acids, and some of the amino acids
most frequently occurring in proteins (entries 1-9, Table 1).
An interesting observation from these data is that the rejection
of the protected amino acids is governed by the character of
the protecting group. For the Fmoc-protected amino acids
without Boc side-chain protection, the rejections are between
40-50%, while those amino acids containing Fmoc as a main-
chain protecting group and Boc as a side-chain protecting group
exhibit rejections within the range of 60-70%. The rejection
increases by +10% per Boc group regardless of the nature of
the amino acids. A probable explanation is that these bulky and
hydrophobic protecting groups impede transport through the
OSN membrane. The relationship between solute structure and
its permeability through the membrane has been studied by
many researchers.33-38 In general the radius of a molecule affects
its diffusivity across the membranesmore bulky molecules pass
through the membrane more slowly and hence exhibit higher
rejection. Steric hindrance caused by larger or more branched
substituents is known to be one of the major factors affecting
rejection of a compound, while interactions between solute and
membrane based on hydrophobicity35-37 or charge also play a
major role in mass transfer across the membrane. This result is
quite different from the results reported in the literature for
nanofiltration of biosynthetic broth, where the rejection of amino
acid strongly depends on the nature of the amino acidsacidic,
basic, hydrophobic, etc.39,40 In aqueous systems, the pH-
dependent ionisation of acidic and basic amino acids apparently
has a larger role in determining rejection.
Further rejection tests on coupling activators and reagents
were performed. Entries 10-14 in Table 1 show that PyBOP
and HBTU activators exhibit relatively high rejection, presum-
ably due to the fact that these activators are bulky salts so that
both steric hindrance and Donnan effects would contribute to
their retention.38 Fortunately, during the coupling reaction these
activators break down into smaller molecules; hence, the
rejection from postreaction solutions is expected to be consider-
ably lower. The coupling reagent DIC showed very low
rejection of 13%, similarly, only 5% of piperidine, the depro-
tection reagent for the Fmoc-deprotection, was rejected (entry
16, Table 1). This is expected since both molecules are small
The results obtained with the Inopor membrane seemed quite
promising. The main drawbacks with using ceramic membranes
are their relatively high price and fragile structure. In this respect
better choices are their counterparts made from polymeric
materials. Unfortunately, the most widely available polymeric
OSN membranes on the market from the StarMem family
manufactured by W.R. Grace are made of polyimide and are
almost instantaneously soluble in DMF. The best commercially
available alternatives from the polymeric OSN membranes
series seemed to be Koch/SelRO MPF-50 (MWCO 700
g·mol-1) (currently not on sale by the manufacturer, but still
(33) Wang, X.-L.; Tsuru, T.; Nakao, S.; Kimura, S. J. Membr. Sci. 1997,
135, 19–32.
(34) Toon, Y.; Lueptow, R. J. Membr. Sci. 2005, 261, 76–86.
(35) Kiso, Y.; Sugiura, Y.; Kitao, T.; Nishimura, K. J. Membr. Sci. 2001,
192, 1–10.
(36) Kiso, Y.; Kon, T.; Kitao, T.; Nishimura, K. J. Membr. Sci. 2001, 182,
205–214.
(37) Kiso, Y.; Nishimura, Y.; Kitao, T.; Nishimura, K. J. Membr. Sci. 2000,
171, 229–237.
(38) Donnan, F. J. Membr. Sci. 1995, 100, 45–55.
(39) Martin-Orue, C.; Bouhallab, S.; Garem, A. J. Membr. Sci. 1998, 142,
225–233.
(40) Tsuru, T.; Nakao, S.; Kimura, S.; Shutou, T. Sep. Sci. Technol. 1994,
29, 971–984.
(41) Carpino, L.; El-Faham, A.; Albericio, F. Tetrahedron Lett. 1994, 35,
2279–2282.
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1319