[discodermolide]

Here is a problem I encountered during the scale-up of a particular compound.
We were using 4-hydrazinobenzoic acid to open a heterocyclic ring, the structure of which is not relevant to this problem. As 4-hydrazinobenzoic acid is carcinogenic the levels of this compound in our final product were set at 0.5 - 0.1ppm or less by QC. We eventually solved it by carrying out two re-crystallisations but on the way we investigated a number of other possibilities.
One was the addition of glucose to form the osazone (compound 1 in attached file), which we could then extract out, however, we did not form the osazone but got something completely different, the structure of which still has not been determined. I suggested compound 2, the NMR spectroscopist compound 3. MS gave a M+ of 314.29 have all the NMR data etc. which I can scan in if you want to see it. We did not manage to get it crystalline for an x-ray!
Has any one here got any other ideas what this compound could be, don’t worry about the stereochemistry.
I’m stumped.

[Dan]

I would also suspect it is the hemiaminal (compound 2) in the first instance. I see no plausible mechanism for the formation of compound 3.

The only other possibility I can think of right now is that you managed to get an Amadori rearrangement to give you a 1-deoxy-1-hydrazinyl-D-fructose, probably in the pyranose form, but your spectroscopist would have spotted the anomeric quaternary centre.

Can you share the NMR?

[discodermolide]

I would also suspect it is the hemiaminal (compound 2) in the first instance. I see no plausible mechanism for the formation of compound 3.

The only other possibility I can think of right now is that you managed to get an Amadori rearrangement to give you a 1-deoxy-1-hydrazinyl-D-fructose, probably in the pyranose form, but your spectroscopist would have spotted the anomeric quaternary centre.

Can you share the NMR?

Thanks Dan. I wasn't aware of the Amadori rearrangement. Give me a while to scan the NMR's

[discodermolide]

I resubmitted the sample under a different number and he came up with structure 4.
The rest of the spectra I will have to post as zip files.
Thanks
Discodermolide

[discodermolide]

Here are the spectra I don't have an A3 scanner so I had to fold them to fit A4, I hope you can use them.

[Dan]

OK, you certainly don't have a ketose, so the fructose derivative I posted is not consistent.

I agree with your assignments H1-H6 (and C1-C6).

I can't see a likely pathway to form compound 4. I think the pyranose hemiaminal compound 2 is consistent - are you sure the doublet you assigned as an OH is not an NH?

How large are the coupling constants? They look quite large to me, though difficult to tell from the pdf, which suggests a pyranose ring (axial-axial coupling ~10 Hz), furanose/furan couplings are typically smaller (up to ~6).

Why don't you think it's compound 2? I guess you are suggestion the 3,5-anydro compound on the basis of the carbon shifts?

[discodermolide]

OK, you certainly don't have a ketose, so the fructose derivative I posted is not consistent.

I agree with your assignments H1-H6 (and C1-C6).

I can't see a likely pathway to form compound 4. I think the pyranose hemiaminal compound 2 is consistent - are you sure the doublet you assigned as an OH is not an NH?

How large are the coupling constants? They look quite large to me, though difficult to tell from the pdf, which suggests a pyranose ring (axial-axial coupling ~10 Hz), furanose/furan couplings are typically smaller (up to ~6).

Why don't you think it's compound 2? I guess you are suggestion the 3,5-anydro compound on the basis of the carbon shifts?

I think it's compound 2. The NMR expert (paid to do structure determination) does not agree.

[Dan]

I'd say the C5 shift is too high for compound 3, but the most obvious problem is the lack of a plausible mechanism. I also think it's probably compound 2.

Can you get an HMBC? That would at least nail the ring size, and might provide some more info on whether the N is exocylcic or endocyclic.

Another option: Hydrogenate the compound over Ni (or maybe Pd/C). If you have compound 3 derived from glucose, you will have 1-deoxynojirimycin after hydrogenation - it's a natural product and a pharmaceutical precursor, easily identified by NMR from the lit. Not only that, but you also have the shortest synthesis of DNJ ever reported.

[discodermolide]

I'd say the C5 shift is too high for compound 3, but the most obvious problem is the lack of a plausible mechanism. I also think it's probably compound 2.

Can you get an HMBC? That would at least nail the ring size, and might provide some more info on whether the N is exocylcic or endocyclic.

Another option: Hydrogenate the compound over Ni (or maybe Pd/C). If you have compound 3 derived from glucose, you will have 1-deoxynojirimycin after hydrogenation - it's a natural product and a pharmaceutical precursor, easily identified by NMR from the lit. Not only that, but you also have the shortest synthesis of DNJ ever reported.

Too late to do anything these days, things have moved on. But this has been nigglig at me for years now, and when I saw your carbohydrate background, well I just had to ask……
I also spoke to George about it when he paid us a visit, as far as I remember he also agreed with us!

[discodermolide]

I'd say the C5 shift is too high for compound 3, but the most obvious problem is the lack of a plausible mechanism. I also think it's probably compound 2.

Can you get an HMBC? That would at least nail the ring size, and might provide some more info on whether the N is exocylcic or endocyclic.

Another option: Hydrogenate the compound over Ni (or maybe Pd/C). If you have compound 3 derived from glucose, you will have 1-deoxynojirimycin after hydrogenation - it's a natural product and a pharmaceutical precursor, easily identified by NMR from the lit. Not only that, but you also have the shortest synthesis of DNJ ever reported.

I thought of that but never got round to it. Feel free to try it if you get time.

[Dan]

Actually, if you did get compound 3, presumably you would have inverted at C5 (D-gluco to L-ido). 1-Deoxy-L-idonojirimycin is not very interesting in comparison to DNJ.

It's not really my are any more, but that's not to say I won't revisit one day. Perhaps one rainy weekend...