Cells in a multicellular organism are surrounded by a matrix of molecules called, appropriately enough, the extracellular matrix (ECM). The ECM is made up of protein and carbohydrate. It anchors the cells and provides structure to the organism. It also provides signaling for mobile cells such as immune system cells.
I was recently talking with a science librarian who's interested in nano/bio stuff, and he told me something that got me thinking. It seems that there's a technology being developed to build many different carbohydrate strands in an array in parallel under optical control, similar to the way an array of DNA strands can be built. The point of this - or at least, one application - is to research the way cells react with the ECM, and perhaps develop new medical sensors.
When we think of biopolymers, it's easy to get stuck thinking about only DNA and protein. In cells, they synthesize each other. Since they are linear chains, they are easy to specify, and can be synthesized by a linear sequence of wet-chemistry steps.
But carbohydrate is also a biopolymer. Perhaps - I haven't researched the details (yet) - it has been less studied in nanotech because a sugar molecule has many p0ssible attachment points for other molecules, so it might be harder to synthesize exactly what you want using wet chemistry.
But if a technology is being developed for synthesizing specific sequences of carbohydrate polymers, then carbohydrates may be usable as building blocks for building nanotech structures. Think of chitin; think of teak. These are very respectable materials.
A molecular manufacturing system based on carbohydrates might start with an extension of whatever approach is being used for synthesizing the carbohydrate arrays. Then, as useful structures were built, it might be possible to mechanically protect and deprotect various sites on the carbohydrate molecules, making the chemistry simpler and more flexible.
One more interesting thing about carbohydrates: because they have multiple attachment points per monomer, it may be reasonable to build molecular structures that are not just linked chain-wise, but crosslinked in 3D. Such structures could be a lot stiffer than linear biopolymers, and stiffness is very desirable at the nanoscale, because it's one of the few properties that doesn't scale well with smaller size. Building 3D crosslinked molecules may be very difficult for wet chemistry, but somewhat easier for mechanically guided chemistry.